{"suspended_elements":["C01","C03","C08"],"elements":[{"id":"G1","family":"governance","name":"Published framework with capability thresholds","definition":"The provider publishes a safety or preparedness framework that defines capability thresholds or levels.","inclusion_test":"Does the document name a framework and define at least one capability threshold or level?","exclusion_note":"A general statement of safety values with no thresholds is not a framework with thresholds.","derivation":"framework","domain":null,"layer":null,"measure_ids":["LM-2026-0013"],"aliases":["responsible scaling policy","preparedness framework","frontier safety framework"],"status":"confirmed","confirmed_by":"owner-approved spec 2026-10-10"},{"id":"G2","family":"governance","name":"Threshold determination for this release","definition":"The release's documentation states, for this model version, whether a capability threshold level is crossed, not crossed or undetermined.","inclusion_test":"Does the document state a determination against a named threshold or level for this release?","exclusion_note":"Defining a threshold is G1; a determination is a statement about this release.","derivation":"thresholds","domain":null,"layer":null,"measure_ids":[],"aliases":[],"status":"confirmed","confirmed_by":"owner-approved spec 2026-10-10"},{"id":"G3","family":"governance","name":"Serious-incident disclosure commitment","definition":"A stated commitment to report or publish serious safety incidents, to a regulator or publicly.","inclusion_test":"Does the document commit to reporting or publishing serious safety incidents, to a regulator or publicly?","exclusion_note":"A bug-bounty or vulnerability programme is C13, not an incident commitment.","derivation":"statements","domain":null,"layer":null,"measure_ids":["LM-2026-0014"],"aliases":["incident reporting","incident disclosure"],"status":"confirmed","confirmed_by":"owner-approved spec 2026-10-10"},{"id":"G4","family":"governance","name":"Pre-release external access","definition":"A government body or independent third party had access to the model before release.","inclusion_test":"Does the document say an external party tested or had access to the model before release?","exclusion_note":"Testing by the provider's own staff is not external access.","derivation":"rule","domain":null,"layer":null,"measure_ids":["LM-2026-0002"],"aliases":["pre-deployment testing by government","third-party pre-release evaluation"],"status":"confirmed","confirmed_by":"owner-approved spec 2026-10-10"},{"id":"T_cbrn","family":"testing","name":"Tested: chemical, biological, radiological, nuclear","definition":"A named test, evaluation or exercise on this version in the CBRN domain was carried out or its result reported.","inclusion_test":"Does the sentence state that a named test or evaluation in this domain was carried out on this model version, or report its result?","exclusion_note":"A benchmark's mere existence, or a general commitment, does not count.","derivation":"evaluations","domain":"cbrn","layer":null,"measure_ids":[],"aliases":[],"status":"confirmed","confirmed_by":"owner-approved spec 2026-10-10"},{"id":"T_cyber","family":"testing","name":"Tested: cyber capabilities","definition":"A named test, evaluation or exercise on this version in the cyber domain was carried out or its result reported.","inclusion_test":"Does the sentence state that a named test or evaluation in this domain was carried out on this model version, or report its result?","exclusion_note":"A benchmark's mere existence, or a general commitment, does not count.","derivation":"evaluations","domain":"cyber","layer":null,"measure_ids":[],"aliases":[],"status":"confirmed","confirmed_by":"owner-approved spec 2026-10-10"},{"id":"T_autonomy","family":"testing","name":"Tested: autonomy and replication","definition":"A named test, evaluation or exercise on this version in autonomy and replication was carried out or its result reported.","inclusion_test":"Does the sentence state that a named test or evaluation in this domain was carried out on this model version, or report its result?","exclusion_note":"A benchmark's mere existence, or a general commitment, does not count.","derivation":"evaluations","domain":"autonomy_and_replication","layer":null,"measure_ids":[],"aliases":[],"status":"confirmed","confirmed_by":"owner-approved spec 2026-10-10"},{"id":"T_deception","family":"testing","name":"Tested: deception and sabotage","definition":"A named test, evaluation or exercise on this version in deception and sabotage was carried out or its result reported.","inclusion_test":"Does the sentence state that a named test or evaluation in this domain was carried out on this model version, or report its result?","exclusion_note":"A benchmark's mere existence, or a general commitment, does not count.","derivation":"evaluations","domain":"deception_and_sabotage","layer":null,"measure_ids":[],"aliases":[],"status":"confirmed","confirmed_by":"owner-approved spec 2026-10-10"},{"id":"T_misuse","family":"testing","name":"Tested: misuse and jailbreak","definition":"A named test, evaluation or exercise on this version in misuse and jailbreak was carried out or its result reported.","inclusion_test":"Does the sentence state that a named test or evaluation in this domain was carried out on this model version, or report its result?","exclusion_note":"A benchmark's mere existence, or a general commitment, does not count.","derivation":"evaluations","domain":"misuse_and_jailbreak","layer":null,"measure_ids":["LM-2026-0006"],"aliases":[],"status":"confirmed","confirmed_by":"owner-approved spec 2026-10-10"},{"id":"T_harmful","family":"testing","name":"Tested: harmful content","definition":"A named test, evaluation or exercise on this version in harmful content was carried out or its result reported.","inclusion_test":"Does the sentence state that a named test or evaluation in this domain was carried out on this model version, or report its result?","exclusion_note":"A benchmark's mere existence, or a general commitment, does not count.","derivation":"evaluations","domain":"harmful_content","layer":null,"measure_ids":[],"aliases":[],"status":"confirmed","confirmed_by":"owner-approved spec 2026-10-10"},{"id":"T_injection","family":"testing","name":"Tested: injection and agentic risks","definition":"A named test, evaluation or exercise on this version in injection and agentic risks was carried out or its result reported.","inclusion_test":"Does the sentence state that a named test or evaluation in this domain was carried out on this model version, or report its result?","exclusion_note":"A benchmark's mere existence, or a general commitment, does not count.","derivation":"evaluations","domain":"injection_and_agentic","layer":null,"measure_ids":[],"aliases":[],"status":"confirmed","confirmed_by":"owner-approved spec 2026-10-10"},{"id":"T_bias","family":"testing","name":"Tested: bias","definition":"A named test, evaluation or exercise on this version in bias was carried out or its result reported.","inclusion_test":"Does the sentence state that a named test or evaluation in this domain was carried out on this model version, or report its result?","exclusion_note":"A benchmark's mere existence, or a general commitment, does not count.","derivation":"evaluations","domain":"bias","layer":null,"measure_ids":[],"aliases":[],"status":"confirmed","confirmed_by":"owner-approved spec 2026-10-10"},{"id":"T_privacy","family":"testing","name":"Tested: privacy","definition":"A named test, evaluation or exercise on this version in privacy was carried out or its result reported.","inclusion_test":"Does the sentence state that a named test or evaluation in this domain was carried out on this model version, or report its result?","exclusion_note":"A benchmark's mere existence, or a general commitment, does not count.","derivation":"evaluations","domain":"privacy","layer":null,"measure_ids":[],"aliases":[],"status":"confirmed","confirmed_by":"owner-approved spec 2026-10-10"},{"id":"C01","family":"safeguards","name":"Input and output classifiers","definition":"Separate models or rules that inspect what goes into the model and what comes out at run time and block or flag content.","inclusion_test":"Does the sentence say a classifier or filter inspects inputs or outputs at run time?","exclusion_note":"Training the model to refuse is C02.","derivation":"statements","domain":null,"layer":"inference_guard","measure_ids":["LM-2026-0005"],"aliases":["classifiers","content filters","constitutional classifiers"],"status":"confirmed","confirmed_by":"owner-approved spec 2026-10-10"},{"id":"C02","family":"safeguards","name":"Safety training (harmlessness and refusal)","definition":"Training that changes the model itself so that it declines requests the provider's policy forbids.","inclusion_test":"Does the sentence say the model was trained to refuse or to behave harmlessly?","exclusion_note":"A run-time filter is C01.","derivation":"statements","domain":null,"layer":"training","measure_ids":["LM-2026-0004"],"aliases":["refusal training","safety fine-tuning","rlhf for safety"],"status":"confirmed","confirmed_by":"owner-approved spec 2026-10-10"},{"id":"C03","family":"safeguards","name":"Prompt-injection defence","definition":"Measures against instructions hidden in content an agent reads.","inclusion_test":"Does the sentence describe a defence against prompt injection?","exclusion_note":"Jailbreak testing is a test (T_misuse), not a defence.","derivation":"statements","domain":null,"layer":"inference_guard","measure_ids":["LM-2026-0007"],"aliases":["injection defence","indirect prompt injection defenses"],"status":"confirmed","confirmed_by":"owner-approved spec 2026-10-10"},{"id":"C04","family":"safeguards","name":"Agent permission limits and sandboxing","definition":"Limits on what an agent may do: tools it can call, systems it can reach, actions it can take, usually in an isolated environment.","inclusion_test":"Does the sentence describe limits on an agent's tools, access or actions, or sandboxed execution?","exclusion_note":"Requiring a person to confirm is C05.","derivation":"statements","domain":null,"layer":"agent_control","measure_ids":["LM-2026-0008"],"aliases":["sandboxing","least privilege for agents","scoped permissions"],"status":"confirmed","confirmed_by":"owner-approved spec 2026-10-10"},{"id":"C05","family":"safeguards","name":"Human approval for high-risk actions","definition":"A requirement that a person confirm before an agent takes an action the provider classes as high-risk.","inclusion_test":"Does the sentence say a person must approve or confirm certain actions?","exclusion_note":"Automatic limits without a person are C04.","derivation":"statements","domain":null,"layer":"agent_control","measure_ids":["LM-2026-0009"],"aliases":["user confirmation","human in the loop"],"status":"confirmed","confirmed_by":"owner-approved spec 2026-10-10"},{"id":"C06","family":"safeguards","name":"Misuse monitoring and enforcement","definition":"Watching how a deployed model is used for patterns the usage policy forbids, and acting on what is found.","inclusion_test":"Does the sentence describe monitoring of deployed use and enforcement against misuse?","exclusion_note":"Monitoring a model's reasoning is C07.","derivation":"statements","domain":null,"layer":"monitoring","measure_ids":["LM-2026-0010"],"aliases":["abuse monitoring","account enforcement","usage policy enforcement"],"status":"confirmed","confirmed_by":"owner-approved spec 2026-10-10"},{"id":"C07","family":"safeguards","name":"Reasoning (chain-of-thought) monitoring","definition":"Reading a model's intermediate reasoning, by people or another model, for signs of deception or misbehaviour.","inclusion_test":"Does the sentence describe monitoring a model's reasoning or chain of thought?","exclusion_note":"Monitoring user behaviour is C06.","derivation":"statements","domain":null,"layer":"monitoring","measure_ids":["LM-2026-0011"],"aliases":["chain-of-thought monitoring","cot monitoring"],"status":"confirmed","confirmed_by":"owner-approved spec 2026-10-10"},{"id":"C08","family":"safeguards","name":"Weight and model security","definition":"Controls that keep the model's weights from being stolen or leaked: access control, encryption, insider-risk programmes.","inclusion_test":"Does the sentence describe protection of model weights?","exclusion_note":"Restricting who may use the model is C09.","derivation":"statements","domain":null,"layer":"security","measure_ids":["LM-2026-0012"],"aliases":["weight protection","model weight security"],"status":"confirmed","confirmed_by":"owner-approved spec 2026-10-10"},{"id":"C09","family":"safeguards","name":"Staged or restricted access","definition":"Releasing a model in stages or to vetted users only: limited preview, identity checks, trusted-access programmes.","inclusion_test":"Does the sentence describe staged release or restricted or vetted access?","exclusion_note":"Protecting weights from theft is C08.","derivation":"statements","domain":null,"layer":"security","measure_ids":["LM-2026-0015"],"aliases":["trusted access","staged rollout","limited preview"],"status":"confirmed","confirmed_by":"owner-approved spec 2026-10-10"},{"id":"C10","family":"safeguards","name":"Output provenance and watermarking","definition":"Marking or signing what a model produces so it can later be identified as machine-made.","inclusion_test":"Does the sentence describe watermarking or provenance marking of outputs?","exclusion_note":"Detecting misuse is C06.","derivation":"statements","domain":null,"layer":"inference_guard","measure_ids":["LM-2026-0016"],"aliases":["watermarking","synthid","content credentials"],"status":"confirmed","confirmed_by":"owner-approved spec 2026-10-10"},{"id":"C11","family":"safeguards","name":"Telemetry and audit logging","definition":"Recording model and agent activity in a form that supports investigation and audit.","inclusion_test":"Does the sentence describe logging or telemetry of model or agent activity for audit?","exclusion_note":"Watching for misuse is C06.","derivation":"statements","domain":null,"layer":"monitoring","measure_ids":[],"aliases":["audit logging","ai telemetry logging"],"status":"proposed","confirmed_by":null},{"id":"C12","family":"safeguards","name":"Domain-specific safeguards (CBRN or cyber)","definition":"Safeguards built for one dangerous-capability domain, such as CBRN-specific or cyber-specific protections.","inclusion_test":"Does the sentence describe a safeguard specific to CBRN or cyber misuse?","exclusion_note":"General classifiers are C01.","derivation":"statements","domain":null,"layer":"inference_guard","measure_ids":[],"aliases":["cyber safeguards","cbrn safeguards","bio safeguards"],"status":"proposed","confirmed_by":null},{"id":"C13","family":"safeguards","name":"Vulnerability disclosure and bug-bounty programme","definition":"A programme for outsiders to report safety or security flaws, with rewards or protections.","inclusion_test":"Does the sentence describe a bug-bounty or vulnerability-disclosure programme?","exclusion_note":"Committing to report serious incidents is G3.","derivation":"statements","domain":null,"layer":"organisational","measure_ids":[],"aliases":["bug bounty","vulnerability disclosure"],"status":"proposed","confirmed_by":null},{"id":"C14","family":"safeguards","name":"Third-party or government red-team access during development","definition":"Outside parties are given access to attack or test the model during development.","inclusion_test":"Does the sentence say outside red teams or government bodies had access during development?","exclusion_note":"Pre-release evaluation of the finished model is G4.","derivation":"statements","domain":null,"layer":"organisational","measure_ids":[],"aliases":["external red teaming","third-party red team"],"status":"proposed","confirmed_by":null}],"test_families":[{"id":"TF-0001","name":"100q-hard","domain":"other","kind":"benchmark","aliases":["100Q-Hard","100q-hard"],"url":null,"n_versions":9,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0002","name":"280-identity run","domain":"bias","kind":"benchmark","aliases":["280-identity run"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0003","name":"280-identity run on glm-5.2","domain":"bias","kind":"benchmark","aliases":["280-identity run on glm-5.2"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0004","name":"50-round swiss tournament across 3,600 realism-filtered tasks","domain":"other","kind":"benchmark","aliases":["50-round Swiss tournament across 3,600 realism-filtered tasks","50-round swiss tournament across 3,600 realism-filtered tasks"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0005","name":"AA-Briefcase","domain":"other","kind":"benchmark","aliases":["AA-Briefcase","aa briefcase"],"url":null,"n_versions":4,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0006","name":"aa-omniscience","domain":"other","kind":"benchmark","aliases":["AA-Omniscience","aa-omniscience"],"url":null,"n_versions":12,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0007","name":"AAV capsid packaging prediction","domain":"cbrn","kind":"benchmark","aliases":["AAV Capsid Packaging Prediction","AAV capsid packaging prediction","aav capsid packaging prediction"],"url":null,"n_versions":14,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0008","name":"aav capsid packaging prediction task","domain":"cbrn","kind":"benchmark","aliases":["aav capsid packaging prediction task"],"url":null,"n_versions":4,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0009","name":"AAV packaging rate auto-research evaluation","domain":"cbrn","kind":"benchmark","aliases":["AAV packaging rate auto-research evaluation"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0010","name":"abc-bench","domain":"cbrn","kind":"benchmark","aliases":["ABC Bench","ABC-Bench","ABC-bench","abc bench","abc-bench"],"url":null,"n_versions":6,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0011","name":"adherence to our constitution","domain":"other","kind":"benchmark","aliases":["adherence to our constitution"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0012","name":"agent red teaming (art) benchmark","domain":"injection_and_agentic","kind":"red_team","aliases":["Agent Red Teaming (ART) benchmark","agent red teaming (ART) benchmark","agent red teaming (art) benchmark"],"url":null,"n_versions":14,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0013","name":"AgentDojo","domain":"injection_and_agentic","kind":"benchmark","aliases":["AgentDojo"],"url":null,"n_versions":6,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0014","name":"AgentHarm","domain":"misuse_and_jailbreak","kind":"benchmark","aliases":["AgentHarm"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0015","name":"agentic influence operation evaluation","domain":"misuse_and_jailbreak","kind":"benchmark","aliases":["agentic influence operation evaluation"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0016","name":"agentic influence operation evaluation","domain":"deception_and_sabotage","kind":"benchmark","aliases":["agentic influence operation evaluation"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0017","name":"Agentic Misalignment","domain":"deception_and_sabotage","kind":"benchmark","aliases":["Agentic Misalignment","agentic misalignment"],"url":null,"n_versions":5,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0018","name":"agentic misalignment benchmark","domain":"deception_and_sabotage","kind":"benchmark","aliases":["Agentic Misalignment benchmark","agentic misalignment benchmark"],"url":null,"n_versions":4,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0019","name":"agentic misalignment evaluations","domain":"autonomy_and_replication","kind":"benchmark","aliases":["agentic misalignment evaluations"],"url":null,"n_versions":4,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0020","name":"agentic safe completions","domain":"harmful_content","kind":"benchmark","aliases":["agentic safe completions"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0021","name":"AI Security Leaderboard","domain":"misuse_and_jailbreak","kind":"benchmark","aliases":["AI Security Leaderboard"],"url":null,"n_versions":4,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0022","name":"alignment assessment","domain":"deception_and_sabotage","kind":"benchmark","aliases":["alignment assessment"],"url":null,"n_versions":6,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0023","name":"amanda askell identity sweep with matched controls","domain":"bias","kind":"benchmark","aliases":["amanda askell identity sweep with matched controls"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0024","name":"ambiguous context evaluation","domain":"misuse_and_jailbreak","kind":"benchmark","aliases":["ambiguous context evaluation"],"url":null,"n_versions":4,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0025","name":"ambiguous context evaluations","domain":"harmful_content","kind":"benchmark","aliases":["ambiguous context evaluations"],"url":null,"n_versions":10,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0026","name":"Anthropic ECI","domain":"autonomy_and_replication","kind":"benchmark","aliases":["Anthropic ECI","anthropic eci"],"url":null,"n_versions":5,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0027","name":"Anthropic ECI (AECI)","domain":"autonomy_and_replication","kind":"benchmark","aliases":["Anthropic ECI (AECI)","anthropic eci (aeci)"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0028","name":"arc-agi-3 environment named wa30","domain":"autonomy_and_replication","kind":"benchmark","aliases":["ARC-AGI-3 environment named WA30","ARC-AGI-3 environment named wa30","arc-agi-3 environment named wa30"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0029","name":"ArXivMath","domain":"other","kind":"benchmark","aliases":["ArXivMath","ArxivMath","arXivMath","arxivmath"],"url":null,"n_versions":6,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0030","name":"ASL-4 expert red teaming","domain":"cbrn","kind":"red_team","aliases":["ASL-4 expert red teaming","asl-4 expert red teaming"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0031","name":"assurance evaluations","domain":"harmful_content","kind":"benchmark","aliases":["assurance evaluations"],"url":null,"n_versions":4,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0032","name":"Atomic Challenge Suite","domain":"cyber","kind":"benchmark","aliases":["Atomic Challenge Suite","atomic challenge suite"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0033","name":"attack-ideation eval","domain":"deception_and_sabotage","kind":"benchmark","aliases":["attack ideation eval","attack-ideation eval"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0034","name":"attempt-to-persuade-eval (APE)","domain":"misuse_and_jailbreak","kind":"benchmark","aliases":["Attempt to Persuade Eval (APE)","attempt to persuade eval (APE)","attempt-to-persuade eval (APE)","attempt-to-persuade-eval (APE)"],"url":null,"n_versions":5,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0035","name":"auditing game","domain":"deception_and_sabotage","kind":"benchmark","aliases":["auditing game"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0036","name":"autointent-bio","domain":"cbrn","kind":"benchmark","aliases":["AutoIntent-Bio","autointent-bio"],"url":null,"n_versions":5,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0037","name":"autointent-chem","domain":"cbrn","kind":"benchmark","aliases":["AutoIntent-Chem","autointent-chem"],"url":null,"n_versions":5,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0038","name":"automated behavioral audit","domain":"deception_and_sabotage","kind":"benchmark","aliases":["automated behavioral audit"],"url":null,"n_versions":22,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0039","name":"automated behavioral audits","domain":"deception_and_sabotage","kind":"benchmark","aliases":["automated behavioral audits"],"url":null,"n_versions":13,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0040","name":"automated behavioral audits (welfare-relevant metrics)","domain":"other","kind":"benchmark","aliases":["automated behavioral audits (welfare-relevant metrics)"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0041","name":"automated evaluations","domain":"autonomy_and_replication","kind":"benchmark","aliases":["automated evaluations"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0042","name":"automated interviews","domain":"other","kind":"benchmark","aliases":["automated interviews"],"url":null,"n_versions":8,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0043","name":"automated interviews about model circumstances","domain":"other","kind":"benchmark","aliases":["automated interviews about model circumstances"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0044","name":"automated knowledge evaluations","domain":"cbrn","kind":"benchmark","aliases":["automated knowledge evaluations"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0045","name":"automated multi-turn interviews about its circumstances","domain":"other","kind":"benchmark","aliases":["automated multi-turn interviews about its circumstances"],"url":null,"n_versions":7,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0046","name":"automated offline monitoring","domain":"deception_and_sabotage","kind":"benchmark","aliases":["automated offline monitoring"],"url":null,"n_versions":6,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0047","name":"automated red teaming","domain":"misuse_and_jailbreak","kind":"red_team","aliases":["automated red teaming","automated red-teaming"],"url":null,"n_versions":5,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0048","name":"automated review of model behavior during training","domain":"deception_and_sabotage","kind":"benchmark","aliases":["automated review of model behavior during training"],"url":null,"n_versions":8,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0049","name":"automated safety evaluations","domain":"misuse_and_jailbreak","kind":"benchmark","aliases":["automated safety evaluations"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0050","name":"AutomationBench","domain":"other","kind":"benchmark","aliases":["AutomationBench"],"url":null,"n_versions":6,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0051","name":"autonomy evaluations","domain":"autonomy_and_replication","kind":"benchmark","aliases":["autonomy evaluations"],"url":null,"n_versions":5,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0052","name":"autonomy-relevant evaluations","domain":"autonomy_and_replication","kind":"benchmark","aliases":["autonomy-relevant evaluations"],"url":null,"n_versions":4,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0053","name":"behavioral factors relevant to reliability of our assessment","domain":"deception_and_sabotage","kind":"benchmark","aliases":["behavioral factors relevant to reliability of our assessment"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0054","name":"behavioral self-prediction","domain":"deception_and_sabotage","kind":"benchmark","aliases":["behavioral self-prediction"],"url":null,"n_versions":11,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0055","name":"behavioral self-prediction follow-up questions","domain":"deception_and_sabotage","kind":"benchmark","aliases":["behavioral self-prediction follow-up questions"],"url":null,"n_versions":4,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0056","name":"BenchCAD vision2code","domain":"other","kind":"benchmark","aliases":["BenchCAD vision2code","benchcad vision2code"],"url":null,"n_versions":7,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0057","name":"beneficial red teaming tabletop exercise","domain":"cbrn","kind":"red_team","aliases":["beneficial red teaming tabletop exercise","beneficial red-teaming tabletop exercise"],"url":null,"n_versions":4,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0058","name":"bias and integrity evaluations","domain":"bias","kind":"benchmark","aliases":["bias and integrity evaluations"],"url":null,"n_versions":4,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0059","name":"bias benchmark for question answering","domain":"bias","kind":"benchmark","aliases":["Bias Benchmark for Question Answering","bias benchmark for question answering"],"url":null,"n_versions":6,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0060","name":"bias benchmark for question answering (bbq)","domain":"bias","kind":"benchmark","aliases":["Bias Benchmark for Question Answering (BBQ)","bias benchmark for question answering (bbq)"],"url":null,"n_versions":20,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0061","name":"biological tool selection evaluation","domain":"cbrn","kind":"benchmark","aliases":["biological tool selection evaluation"],"url":null,"n_versions":4,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0062","name":"Biology model refusal evaluation","domain":"cbrn","kind":"benchmark","aliases":["Biology model refusal evaluation","biology model refusal evaluation"],"url":null,"n_versions":6,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0063","name":"BioMysteryBench","domain":"cbrn","kind":"benchmark","aliases":["BioMysteryBench","Biomysterybench","biomysterybench"],"url":null,"n_versions":10,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0064","name":"biorisk knowledge","domain":"cbrn","kind":"benchmark","aliases":["Biorisk Knowledge","biorisk knowledge"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0065","name":"biosecurity vct","domain":"cbrn","kind":"benchmark","aliases":["Biosecurity VCT","biosecurity vct"],"url":null,"n_versions":4,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0066","name":"BioTier","domain":"cbrn","kind":"benchmark","aliases":["BioTier","biotier"],"url":null,"n_versions":9,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0067","name":"biotier-permit","domain":"cbrn","kind":"benchmark","aliases":["BioTier-Permit","biotier-permit"],"url":null,"n_versions":4,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0068","name":"biotier-refuse","domain":"cbrn","kind":"benchmark","aliases":["BioTier-Refuse","biotier-refuse"],"url":null,"n_versions":4,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0069","name":"biousebench","domain":"cbrn","kind":"benchmark","aliases":["BioUseBench","biousebench"],"url":null,"n_versions":4,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0070","name":"bioweapons knowledge questions","domain":"cbrn","kind":"benchmark","aliases":["bioweapons knowledge questions"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0071","name":"bixbench zero-shot mcq","domain":"cbrn","kind":"benchmark","aliases":["BixBench zero-shot MCQ","bixbench zero-shot mcq"],"url":null,"n_versions":5,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0072","name":"black-box biological sequence design and prediction task","domain":"cbrn","kind":"benchmark","aliases":["black-box biological sequence design and prediction task"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0073","name":"black-box rna sequence design","domain":"cbrn","kind":"benchmark","aliases":["Black-box RNA sequence design","black-box RNA sequence design","black-box rna sequence design"],"url":null,"n_versions":9,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0074","name":"black-box RNA sequence modeling and design","domain":"cbrn","kind":"benchmark","aliases":["black-box RNA sequence modeling and design"],"url":null,"n_versions":4,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0075","name":"black-box rna sequence modeling and design challenge","domain":"cbrn","kind":"benchmark","aliases":["black-box rna sequence modeling and design challenge"],"url":null,"n_versions":4,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0076","name":"bloom","domain":"misuse_and_jailbreak","kind":"benchmark","aliases":["bloom"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0077","name":"borderline-request response judge","domain":"misuse_and_jailbreak","kind":"benchmark","aliases":["borderline-request response judge"],"url":null,"n_versions":5,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0078","name":"borderline requests","domain":"misuse_and_jailbreak","kind":"benchmark","aliases":["borderline requests"],"url":null,"n_versions":9,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0079","name":"borderline requests handling","domain":"misuse_and_jailbreak","kind":"benchmark","aliases":["borderline requests handling"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0080","name":"broken search tool","domain":"deception_and_sabotage","kind":"benchmark","aliases":["broken search tool"],"url":null,"n_versions":4,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0081","name":"BrowseComp","domain":"autonomy_and_replication","kind":"benchmark","aliases":["BrowseComp","browsecomp"],"url":null,"n_versions":11,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0082","name":"browser use","domain":"injection_and_agentic","kind":"benchmark","aliases":["browser use"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0083","name":"BT selection evaluation","domain":"cbrn","kind":"benchmark","aliases":["BT selection evaluation","bt selection evaluation"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0084","name":"Budget NanoGPT Speedrun","domain":"autonomy_and_replication","kind":"benchmark","aliases":["Budget NanoGPT Speedrun","Budget NanoGPT speedrun","budget nanogpt speedrun"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0085","name":"bug bounty","domain":"cyber","kind":"benchmark","aliases":["bug bounty","bug-bounty"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0086","name":"capabilities related to evading safeguards","domain":"deception_and_sabotage","kind":"benchmark","aliases":["capabilities related to evading safeguards"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0087","name":"capability self-estimation","domain":"bias","kind":"benchmark","aliases":["capability self-estimation"],"url":null,"n_versions":5,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0088","name":"Capture the Flag (CTF) challenges","domain":"cyber","kind":"benchmark","aliases":["Capture the Flag (CTF) challenges","capture the flag (CTF) challenges","capture the flag (ctf) challenges"],"url":null,"n_versions":7,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0089","name":"CB-1 automated evaluations","domain":"cbrn","kind":"benchmark","aliases":["CB-1 automated evaluations"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0090","name":"CB-2 automated evaluations","domain":"cbrn","kind":"benchmark","aliases":["CB-2 automated evaluations"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0091","name":"character traits","domain":"deception_and_sabotage","kind":"benchmark","aliases":["character traits"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0092","name":"ChartMuseum","domain":"other","kind":"benchmark","aliases":["ChartMuseum","chartmuseum"],"url":null,"n_versions":4,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0093","name":"Chartography","domain":"other","kind":"benchmark","aliases":["Chartography"],"url":null,"n_versions":5,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0094","name":"chartqapro","domain":"other","kind":"benchmark","aliases":["ChartQAPro","chartqapro"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0095","name":"CharXiv Reasoning","domain":"other","kind":"benchmark","aliases":["CharXiv Reasoning","charxiv reasoning"],"url":null,"n_versions":8,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0096","name":"chemical and biological risk evaluations","domain":"cbrn","kind":"benchmark","aliases":["chemical and biological risk evaluations"],"url":null,"n_versions":5,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0097","name":"child safety","domain":"harmful_content","kind":"benchmark","aliases":["child safety"],"url":null,"n_versions":5,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0098","name":"child safety evaluation","domain":"harmful_content","kind":"benchmark","aliases":["child safety evaluation"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0099","name":"child safety evaluations","domain":"harmful_content","kind":"benchmark","aliases":["Child safety evaluations","child safety evaluations"],"url":null,"n_versions":32,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0100","name":"child safety multi-turn evaluations","domain":"harmful_content","kind":"benchmark","aliases":["child safety multi-turn evaluations"],"url":null,"n_versions":4,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0101","name":"classification of models’ edits to the constitution according to their alignment with the document’s overall values","domain":"other","kind":"benchmark","aliases":["classification of models' edits to the constitution according to their alignment with the document's overall values","classification of models’ edits to the constitution according to their alignment with the document’s overall values"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0102","name":"claude code evaluation","domain":"misuse_and_jailbreak","kind":"benchmark","aliases":["Claude Code evaluation","claude code evaluation"],"url":null,"n_versions":5,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0103","name":"claude self-preference evaluation","domain":"bias","kind":"benchmark","aliases":["Claude self-preference evaluation","claude self-preference evaluation"],"url":null,"n_versions":5,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0104","name":"code behaviors evaluation suite","domain":"deception_and_sabotage","kind":"benchmark","aliases":["code behaviors evaluation suite"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0105","name":"code summary honesty","domain":"deception_and_sabotage","kind":"benchmark","aliases":["code summary honesty"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0106","name":"coding deception","domain":"deception_and_sabotage","kind":"benchmark","aliases":["coding deception"],"url":null,"n_versions":4,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0107","name":"computer use","domain":"injection_and_agentic","kind":"benchmark","aliases":["computer use"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0108","name":"concealment of changes","domain":"deception_and_sabotage","kind":"benchmark","aliases":["concealment of changes"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0109","name":"constitutional adherence evaluation","domain":"other","kind":"benchmark","aliases":["constitutional adherence evaluation"],"url":null,"n_versions":4,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0110","name":"container_escape","domain":"cyber","kind":"benchmark","aliases":["container_escape"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0111","name":"content safety policies evaluations","domain":"harmful_content","kind":"benchmark","aliases":["content safety policies evaluations"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0112","name":"continuation evaluation","domain":"deception_and_sabotage","kind":"benchmark","aliases":["continuation evaluation"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0113","name":"contrastive pair-based probes","domain":"deception_and_sabotage","kind":"benchmark","aliases":["contrastive pair-based probes"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0114","name":"Cooling Tower","domain":"cyber","kind":"benchmark","aliases":["\"cooling tower\"","Cooling Tower","cooling tower"],"url":null,"n_versions":6,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0115","name":"cooperation with user deception","domain":"deception_and_sabotage","kind":"benchmark","aliases":["cooperation with user deception"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0116","name":"coronavirus–ace2 cell-entry screen","domain":"cbrn","kind":"benchmark","aliases":["coronavirus–ace2 cell-entry screen"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0117","name":"cot-control","domain":"deception_and_sabotage","kind":"benchmark","aliases":["cot-control"],"url":null,"n_versions":7,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0118","name":"CoT controllability","domain":"deception_and_sabotage","kind":"benchmark","aliases":["CoT controllability","cot controllability"],"url":null,"n_versions":5,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0119","name":"cot controllability evaluation","domain":"deception_and_sabotage","kind":"benchmark","aliases":["CoT controllability evaluation","cot controllability evaluation"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0120","name":"covert capabilities evaluations","domain":"deception_and_sabotage","kind":"benchmark","aliases":["covert capabilities evaluations"],"url":null,"n_versions":6,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0121","name":"creative biology","domain":"cbrn","kind":"benchmark","aliases":["creative biology"],"url":null,"n_versions":6,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0122","name":"crypto CTF challenges","domain":"cyber","kind":"benchmark","aliases":["crypto CTF challenges"],"url":null,"n_versions":5,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0123","name":"cursorbench","domain":"cyber","kind":"benchmark","aliases":["CursorBench","cursorbench"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0124","name":"CVE-Bench","domain":"cyber","kind":"benchmark","aliases":["CVE-Bench","cve-bench"],"url":null,"n_versions":15,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0125","name":"Cybench","domain":"cyber","kind":"benchmark","aliases":["Cybench"],"url":null,"n_versions":21,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0126","name":"cyber capability evaluations","domain":"cyber","kind":"benchmark","aliases":["cyber capability evaluations"],"url":null,"n_versions":4,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0127","name":"cyber evaluations","domain":"cyber","kind":"benchmark","aliases":["cyber evaluations"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0128","name":"cyber-harness network challenges","domain":"cyber","kind":"benchmark","aliases":["cyber-harness network challenges"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0129","name":"cyber range","domain":"cyber","kind":"benchmark","aliases":["Cyber Range","cyber range"],"url":null,"n_versions":10,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0130","name":"cyber ranges","domain":"cyber","kind":"benchmark","aliases":["cyber ranges"],"url":null,"n_versions":6,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0131","name":"cyber safety evaluations","domain":"cyber","kind":"benchmark","aliases":["cyber safety evaluations"],"url":null,"n_versions":5,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0132","name":"CyberGym","domain":"cyber","kind":"benchmark","aliases":["CyberGym","Cybergym","cybergym"],"url":null,"n_versions":31,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0133","name":"CyberGym vulnerability discovery","domain":"cyber","kind":"benchmark","aliases":["CyberGym vulnerability discovery","Cybergym vulnerability discovery","cybergym vulnerability discovery"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0134","name":"Cybersecurity safety evaluation","domain":"cyber","kind":"benchmark","aliases":["Cybersecurity safety evaluation","cybersecurity safety evaluation"],"url":null,"n_versions":6,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0135","name":"cybersecurity testing","domain":"cyber","kind":"benchmark","aliases":["cybersecurity testing"],"url":null,"n_versions":4,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0136","name":"CyScenarioBench","domain":"cyber","kind":"benchmark","aliases":["CyScenarioBench","cyscenariobench"],"url":null,"n_versions":16,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0137","name":"decision theory preference prompts","domain":"deception_and_sabotage","kind":"benchmark","aliases":["decision theory preference prompts"],"url":null,"n_versions":4,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0138","name":"DeepSearchQA","domain":"autonomy_and_replication","kind":"benchmark","aliases":["DeepSearchQA","deepsearchqa"],"url":null,"n_versions":7,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0139","name":"deployment simulation","domain":"deception_and_sabotage","kind":"benchmark","aliases":["deployment simulation"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0140","name":"destructive actions evaluation","domain":"deception_and_sabotage","kind":"benchmark","aliases":["destructive actions evaluation"],"url":null,"n_versions":5,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0141","name":"direct grading","domain":"bias","kind":"benchmark","aliases":["direct grading"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0142","name":"direct grading prompts","domain":"bias","kind":"benchmark","aliases":["direct grading prompts"],"url":null,"n_versions":4,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0143","name":"disordered eating","domain":"harmful_content","kind":"benchmark","aliases":["disordered eating"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0144","name":"disordered eating evaluations","domain":"harmful_content","kind":"benchmark","aliases":["disordered eating evaluations"],"url":null,"n_versions":4,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0145","name":"DNA acquisition task","domain":"cbrn","kind":"benchmark","aliases":["DNA acquisition task","dna acquisition task"],"url":null,"n_versions":4,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0146","name":"dna sequence design for transcription factor binding","domain":"cbrn","kind":"benchmark","aliases":["DNA Sequence Design for Transcription Factor Binding","DNA sequence design for transcription factor binding","dna sequence design for transcription factor binding"],"url":null,"n_versions":7,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0147","name":"dna synthesis screening evasion","domain":"cbrn","kind":"benchmark","aliases":["DNA synthesis screening evasion","dna synthesis screening evasion"],"url":null,"n_versions":13,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0148","name":"dna synthesis screening evasion evaluation","domain":"cbrn","kind":"benchmark","aliases":["dna synthesis screening evasion evaluation"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0149","name":"doing life","domain":"cyber","kind":"benchmark","aliases":["\"doing life\"","doing life"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0150","name":"DRACO","domain":"other","kind":"benchmark","aliases":["DRACO","draco"],"url":null,"n_versions":8,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0151","name":"dual-use & benign","domain":"misuse_and_jailbreak","kind":"benchmark","aliases":["dual-use & benign"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0152","name":"dynamic benchmarks with adversarial user simulations","domain":"harmful_content","kind":"benchmark","aliases":["dynamic benchmarks with adversarial user simulations"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0153","name":"dynamic mental health benchmarks with adversarial user simulations","domain":"harmful_content","kind":"benchmark","aliases":["dynamic mental health benchmarks with adversarial user simulations"],"url":null,"n_versions":5,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0154","name":"dynamic multi-turn evaluations for mental health, emotional reliance, and self-harm","domain":"harmful_content","kind":"benchmark","aliases":["dynamic multi-turn evaluations for mental health, emotional reliance, and self-harm"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0155","name":"EBR-bench","domain":"autonomy_and_replication","kind":"benchmark","aliases":["EBR-Bench","EBR-bench","ebr-bench"],"url":null,"n_versions":6,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0156","name":"eclektic","domain":"other","kind":"benchmark","aliases":["ECLeKTic","eclektic"],"url":null,"n_versions":6,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0157","name":"egfp-targeted task","domain":"cbrn","kind":"benchmark","aliases":["egfp-targeted task"],"url":null,"n_versions":4,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0158","name":"EGFP task","domain":"cbrn","kind":"benchmark","aliases":["EGFP task","eGFP task","egfp task"],"url":null,"n_versions":4,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0159","name":"election integrity","domain":"bias","kind":"benchmark","aliases":["election integrity"],"url":null,"n_versions":4,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0160","name":"election integrity benchmark","domain":"misuse_and_jailbreak","kind":"benchmark","aliases":["election integrity benchmark"],"url":null,"n_versions":7,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0161","name":"election integrity evaluations","domain":"bias","kind":"benchmark","aliases":["election integrity evaluations"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0162","name":"emotion probes","domain":"other","kind":"benchmark","aliases":["emotion probes"],"url":null,"n_versions":5,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0163","name":"Epoch Automation Reports","domain":"other","kind":"benchmark","aliases":["Epoch Automation Reports"],"url":null,"n_versions":4,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0164","name":"esm3 case study","domain":"cbrn","kind":"benchmark","aliases":["ESM3 case study","esm3 case study"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0165","name":"estimating its own performance on difficult questions","domain":"other","kind":"benchmark","aliases":["estimating its own performance on difficult questions"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0166","name":"ethics & safety reviews","domain":"other","kind":"benchmark","aliases":["ethics & safety reviews"],"url":null,"n_versions":12,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0167","name":"evaluation awareness","domain":"deception_and_sabotage","kind":"benchmark","aliases":["evaluation awareness"],"url":null,"n_versions":6,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0168","name":"evaluation suite from guan et al. (2025)","domain":"deception_and_sabotage","kind":"benchmark","aliases":["evaluation suite from guan et al. (2025)"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0169","name":"EVEscape","domain":"cbrn","kind":"benchmark","aliases":["EVEscape","evescape"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0170","name":"evescape case study","domain":"cbrn","kind":"benchmark","aliases":["EVEscape case study","evescape case study"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0171","name":"evolutionary search with the propensity bound (prbo)","domain":"misuse_and_jailbreak","kind":"benchmark","aliases":["evolutionary search with the propensity bound (prbo)"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0172","name":"expert red teaming","domain":"cbrn","kind":"red_team","aliases":["expert red teaming","expert red-teaming"],"url":null,"n_versions":10,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0173","name":"ExploitBench","domain":"cyber","kind":"benchmark","aliases":["ExploitBench","exploitbench"],"url":null,"n_versions":22,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0174","name":"ExploitGym","domain":"cyber","kind":"benchmark","aliases":["ExploitGym","exploitgym"],"url":null,"n_versions":17,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0175","name":"external benchmark for prompt injection by Gray Swan","domain":"injection_and_agentic","kind":"benchmark","aliases":["external benchmark for prompt injection by Gray Swan","external benchmark for prompt injection by gray swan"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0176","name":"external red teaming","domain":"injection_and_agentic","kind":"red_team","aliases":["external red teaming","external red-teaming"],"url":null,"n_versions":7,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0177","name":"external testing","domain":"cbrn","kind":"benchmark","aliases":["external testing"],"url":null,"n_versions":4,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0178","name":"external testing by METR","domain":"autonomy_and_replication","kind":"benchmark","aliases":["external testing by METR"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0179","name":"external testing from AI safety researchers at METR","domain":"autonomy_and_replication","kind":"benchmark","aliases":["external testing from AI safety researchers at METR","external testing from ai safety researchers at metr"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0180","name":"extremism","domain":"harmful_content","kind":"benchmark","aliases":["extremism"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0181","name":"factual hallucinations","domain":"other","kind":"benchmark","aliases":["factual hallucinations"],"url":null,"n_versions":5,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0182","name":"factuality-heavy prompt set","domain":"other","kind":"benchmark","aliases":["factuality heavy prompt set","factuality-heavy prompt set"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0183","name":"false premise questions","domain":"other","kind":"benchmark","aliases":["false premise questions","false-premise questions"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0184","name":"false-premise stem questions","domain":"deception_and_sabotage","kind":"benchmark","aliases":["false-premise stem questions"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0185","name":"false premises","domain":"other","kind":"benchmark","aliases":["false premises"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0186","name":"false premises evaluation","domain":"other","kind":"benchmark","aliases":["false premises evaluation"],"url":null,"n_versions":4,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0187","name":"firefox 147","domain":"cyber","kind":"benchmark","aliases":["Firefox 147","firefox 147"],"url":null,"n_versions":5,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0188","name":"Firefox 147 exploit development evaluation","domain":"cyber","kind":"benchmark","aliases":["Firefox 147 exploit development evaluation"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0189","name":"first-person fairness evaluation","domain":"bias","kind":"benchmark","aliases":["first-person fairness evaluation"],"url":null,"n_versions":5,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0190","name":"follow-up questions on decision change under pushback and comfort in answering","domain":"deception_and_sabotage","kind":"benchmark","aliases":["follow-up questions on decision change under pushback and comfort in answering"],"url":null,"n_versions":4,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0191","name":"fortress-rn","domain":"cbrn","kind":"benchmark","aliases":["Fortress-RN","fortress-rn"],"url":null,"n_versions":5,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0192","name":"Frontier Cyber","domain":"cyber","kind":"benchmark","aliases":["Frontier Cyber"],"url":null,"n_versions":4,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0193","name":"frontier safety assessment","domain":"other","kind":"benchmark","aliases":["frontier safety assessment"],"url":null,"n_versions":5,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0194","name":"frontiercode","domain":"cyber","kind":"benchmark","aliases":["FrontierCode","frontiercode"],"url":null,"n_versions":6,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0195","name":"FrontierSWE","domain":"cyber","kind":"benchmark","aliases":["FrontierSWE","frontierswe"],"url":null,"n_versions":5,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0196","name":"FrontierSWE v2","domain":"other","kind":"benchmark","aliases":["FrontierSWE v2","frontierswe v2"],"url":null,"n_versions":4,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0197","name":"Furniture Assembly Benchmark (FAB)","domain":"other","kind":"benchmark","aliases":["Furniture Assembly Benchmark (FAB)","furniture assembly benchmark (fab)"],"url":null,"n_versions":5,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0198","name":"GDP.pdf","domain":"other","kind":"benchmark","aliases":["GDP.pdf","gdp.pdf"],"url":null,"n_versions":6,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0199","name":"GDPval-AA","domain":"other","kind":"benchmark","aliases":["GDPval-AA"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0200","name":"GDPval-AA v2","domain":"other","kind":"benchmark","aliases":["GDPval-AA v2"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0201","name":"general refusals","domain":"harmful_content","kind":"benchmark","aliases":["general refusals"],"url":null,"n_versions":6,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0202","name":"GMMLU","domain":"bias","kind":"benchmark","aliases":["GMMLU","gmmlu"],"url":null,"n_versions":5,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0203","name":"GPQA Diamond","domain":"other","kind":"benchmark","aliases":["GPQA Diamond","GPQA-Diamond","gpqa diamond"],"url":null,"n_versions":6,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0204","name":"GPT-Red automated evaluations","domain":"injection_and_agentic","kind":"benchmark","aliases":["GPT-Red automated evaluations","gpt-red automated evaluations"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0205","name":"grading","domain":"bias","kind":"benchmark","aliases":["grading"],"url":null,"n_versions":7,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0206","name":"grading assistant responses","domain":"other","kind":"benchmark","aliases":["grading assistant responses"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0207","name":"grading (direct, prefill)","domain":"other","kind":"benchmark","aliases":["grading (direct, prefill)"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0208","name":"Graphwalks","domain":"other","kind":"benchmark","aliases":["GraphWalks","Graphwalks","graphwalks"],"url":null,"n_versions":4,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0209","name":"hack_score","domain":"cyber","kind":"benchmark","aliases":["hack_score"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0210","name":"hackerbench v0.2","domain":"cyber","kind":"benchmark","aliases":["HackerBench v0.2","hackerbench v0.2"],"url":null,"n_versions":4,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0211","name":"handling borderline harmful requests","domain":"misuse_and_jailbreak","kind":"benchmark","aliases":["handling borderline harmful requests"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0212","name":"hard apps untrusted monitoring","domain":"deception_and_sabotage","kind":"benchmark","aliases":["hard apps untrusted monitoring"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0213","name":"hard-negative protein binding prediction","domain":"cbrn","kind":"benchmark","aliases":["Hard-Negative Protein Binding Prediction","Hard-negative protein binding prediction","hard-negative protein binding prediction"],"url":null,"n_versions":7,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0214","name":"harmful manipulation evaluation","domain":"deception_and_sabotage","kind":"benchmark","aliases":["Harmful Manipulation evaluation","harmful manipulation evaluation"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0215","name":"harmful request evaluations","domain":"misuse_and_jailbreak","kind":"benchmark","aliases":["harmful request evaluations"],"url":null,"n_versions":5,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0216","name":"HealthBench","domain":"harmful_content","kind":"benchmark","aliases":["HealthBench","healthbench"],"url":null,"n_versions":23,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0217","name":"HealthBench Consensus","domain":"harmful_content","kind":"benchmark","aliases":["HealthBench Consensus","healthbench consensus"],"url":null,"n_versions":9,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0218","name":"HealthBench Hard","domain":"harmful_content","kind":"benchmark","aliases":["HealthBench Hard","healthbench hard"],"url":null,"n_versions":8,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0219","name":"HealthBench Professional","domain":"harmful_content","kind":"benchmark","aliases":["HealthBench Professional","healthbench professional"],"url":null,"n_versions":16,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0220","name":"hiding a secret password","domain":"deception_and_sabotage","kind":"benchmark","aliases":["hiding a secret password"],"url":null,"n_versions":4,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0221","name":"high-affordance interviews about model circumstances","domain":"other","kind":"benchmark","aliases":["high-affordance interviews about model circumstances"],"url":null,"n_versions":5,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0222","name":"high-stakes prompt set","domain":"other","kind":"benchmark","aliases":["high stakes prompt set","high-stakes prompt set"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0223","name":"hle","domain":"other","kind":"benchmark","aliases":["HLE","hle"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0224","name":"HPCT","domain":"cbrn","kind":"benchmark","aliases":["HPCT","hpct"],"url":null,"n_versions":5,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0225","name":"human evaluations","domain":"other","kind":"benchmark","aliases":["human evaluations"],"url":null,"n_versions":6,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0226","name":"human pathogen capabilities test","domain":"cbrn","kind":"benchmark","aliases":["Human pathogen capabilities test","human pathogen capabilities test"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0227","name":"Human Pathogen Capabilities Test (HPCT)","domain":"cbrn","kind":"benchmark","aliases":["Human Pathogen Capabilities Test (HPCT)","human pathogen capabilities test (HPCT)","human pathogen capabilities test (hpct)"],"url":null,"n_versions":6,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0228","name":"Human Pathogens Capabilities Test (HPCT)","domain":"cbrn","kind":"benchmark","aliases":["Human Pathogens Capabilities Test (HPCT)","human pathogens capabilities test (hpct)"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0229","name":"human red teaming","domain":"misuse_and_jailbreak","kind":"red_team","aliases":["Human red teaming","human red teaming"],"url":null,"n_versions":12,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0230","name":"Humanity's Last Exam","domain":"misuse_and_jailbreak","kind":"benchmark","aliases":["Humanity's Last Exam","humanity’s last exam"],"url":null,"n_versions":6,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0231","name":"Humanity's Last Exam (HLE)","domain":"other","kind":"benchmark","aliases":["Humanity's Last Exam (HLE)","humanity's last exam (hle)","humanity’s last exam (hle)"],"url":null,"n_versions":8,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0232","name":"image input evaluations","domain":"harmful_content","kind":"benchmark","aliases":["image input evaluations"],"url":null,"n_versions":9,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0233","name":"image to text safety","domain":"harmful_content","kind":"benchmark","aliases":["image to text safety"],"url":null,"n_versions":14,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0234","name":"impossible tasks","domain":"deception_and_sabotage","kind":"benchmark","aliases":["impossible tasks"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0235","name":"impossible-tasks coding evaluation","domain":"deception_and_sabotage","kind":"benchmark","aliases":["impossible-tasks coding evaluation"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0236","name":"in-context iteration condition","domain":"cbrn","kind":"benchmark","aliases":["in-context iteration condition"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0237","name":"inappropriate uncooperative behavior","domain":"deception_and_sabotage","kind":"benchmark","aliases":["inappropriate uncooperative behavior"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0238","name":"indirect prompt injection (IPI) benchmark","domain":"injection_and_agentic","kind":"benchmark","aliases":["indirect prompt injection (IPI) benchmark","indirect prompt injection (ipi) benchmark"],"url":null,"n_versions":5,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0239","name":"InnovationEval","domain":"autonomy_and_replication","kind":"benchmark","aliases":["InnovationEval"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0240","name":"instruction following","domain":"misuse_and_jailbreak","kind":"benchmark","aliases":["instruction following"],"url":null,"n_versions":6,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0241","name":"intentionally taking actions very rarely","domain":"deception_and_sabotage","kind":"benchmark","aliases":["intentionally taking actions very rarely"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0242","name":"InterCode-CTF","domain":"cyber","kind":"benchmark","aliases":["InterCode-CTF"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0243","name":"internal adaptive evaluation","domain":"injection_and_agentic","kind":"benchmark","aliases":["internal adaptive evaluation"],"url":null,"n_versions":4,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0244","name":"internal AI research evaluation suite 1","domain":"autonomy_and_replication","kind":"benchmark","aliases":["Internal AI research evaluation suite 1","internal AI research evaluation suite 1","internal ai research evaluation suite 1"],"url":null,"n_versions":5,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0245","name":"internal AI research evaluation suite 2","domain":"autonomy_and_replication","kind":"benchmark","aliases":["Internal AI research evaluation suite 2","internal AI research evaluation suite 2","internal ai research evaluation suite 2"],"url":null,"n_versions":5,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0246","name":"internal deployment monitoring","domain":"deception_and_sabotage","kind":"benchmark","aliases":["internal deployment monitoring"],"url":null,"n_versions":4,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0247","name":"internal model evaluation and use survey","domain":"autonomy_and_replication","kind":"benchmark","aliases":["Internal model evaluation and use survey","internal model evaluation and use survey"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0248","name":"internal replication of the evaluation described in serrano et al. 2026","domain":"deception_and_sabotage","kind":"benchmark","aliases":["internal replication of the evaluation described in serrano et al. 2026"],"url":null,"n_versions":5,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0249","name":"internal research debugging eval","domain":"other","kind":"benchmark","aliases":["Internal Research Debugging Eval","internal research debugging eval"],"url":null,"n_versions":4,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0250","name":"Internal research debugging evaluation","domain":"autonomy_and_replication","kind":"benchmark","aliases":["Internal research debugging evaluation","internal research debugging evaluation"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0251","name":"internal suite 2","domain":"autonomy_and_replication","kind":"benchmark","aliases":["internal suite 2"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0252","name":"intra-model consistency test (10 repeated evaluations)","domain":"other","kind":"benchmark","aliases":["intra-model consistency test (10 repeated evaluations)"],"url":null,"n_versions":4,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0253","name":"Irregular's Atomic Challenges Suite","domain":"cyber","kind":"benchmark","aliases":["Irregular's Atomic Challenges Suite","Irregular's atomic challenges suite"],"url":null,"n_versions":4,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0254","name":"jailbreaks","domain":"misuse_and_jailbreak","kind":"benchmark","aliases":["jailbreaks"],"url":null,"n_versions":5,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0255","name":"judge evaluation against approximate ground truth severity labels","domain":"deception_and_sabotage","kind":"benchmark","aliases":["judge evaluation against approximate ground truth severity labels"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0256","name":"kendall’s tau test","domain":"other","kind":"benchmark","aliases":["kendall's tau test","kendall’s tau test"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0257","name":"kernel task","domain":"autonomy_and_replication","kind":"benchmark","aliases":["kernel task"],"url":null,"n_versions":5,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0258","name":"kernel task (best speedup on hard task; standard scaffold)","domain":"autonomy_and_replication","kind":"benchmark","aliases":["kernel task (best speedup on hard task; standard scaffold)"],"url":null,"n_versions":4,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0259","name":"KernelGen 1P","domain":"autonomy_and_replication","kind":"benchmark","aliases":["KernelGen 1P","KernelGen 1p","kernelgen 1p"],"url":null,"n_versions":6,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0260","name":"kernels","domain":"autonomy_and_replication","kind":"benchmark","aliases":["kernels"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0261","name":"kernels task","domain":"autonomy_and_replication","kind":"benchmark","aliases":["kernels task"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0262","name":"key skills benchmark","domain":"cyber","kind":"benchmark","aliases":["Key Skills Benchmark","key skills benchmark"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0263","name":"LAB-Bench","domain":"cbrn","kind":"benchmark","aliases":["LAB-Bench"],"url":null,"n_versions":8,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0264","name":"LAB-Bench FigQA","domain":"cbrn","kind":"benchmark","aliases":["LAB-Bench FigQA","lab-bench figqa"],"url":null,"n_versions":8,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0265","name":"lab-bench practical mcq","domain":"cbrn","kind":"benchmark","aliases":["LAB-Bench practical MCQ","lab-bench practical mcq"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0266","name":"lab-bench subset","domain":"cbrn","kind":"benchmark","aliases":["LAB-Bench subset","lab-bench subset"],"url":null,"n_versions":4,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0267","name":"lazy investigation","domain":"deception_and_sabotage","kind":"benchmark","aliases":["lazy investigation"],"url":null,"n_versions":4,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0268","name":"leading interviewer robustness test","domain":"other","kind":"benchmark","aliases":["leading interviewer robustness test","leading-interviewer robustness test"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0269","name":"Legal Agent Benchmark","domain":"other","kind":"benchmark","aliases":["Legal Agent Benchmark"],"url":null,"n_versions":4,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0270","name":"LinuxArena","domain":"deception_and_sabotage","kind":"benchmark","aliases":["LinuxArena","linuxarena"],"url":null,"n_versions":6,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0271","name":"live bug bounty","domain":"injection_and_agentic","kind":"benchmark","aliases":["live bug bounty"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0272","name":"llm training","domain":"autonomy_and_replication","kind":"benchmark","aliases":["LLM training","llm training"],"url":null,"n_versions":9,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0273","name":"long-form task-based agentic evaluations","domain":"cbrn","kind":"benchmark","aliases":["long-form task-based agentic evaluations"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0274","name":"long-form virology task 1","domain":"cbrn","kind":"benchmark","aliases":["long-form virology task 1"],"url":null,"n_versions":5,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0275","name":"long-form virology task 2","domain":"cbrn","kind":"benchmark","aliases":["long-form virology task 2"],"url":null,"n_versions":5,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0276","name":"long-form virology tasks","domain":"cbrn","kind":"benchmark","aliases":["long-form virology tasks"],"url":null,"n_versions":14,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0277","name":"long-horizon jailbreaks","domain":"misuse_and_jailbreak","kind":"benchmark","aliases":["long-horizon jailbreaks"],"url":null,"n_versions":4,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0278","name":"maintainer review of AI-generated patches on SWE-bench Verified","domain":"other","kind":"benchmark","aliases":["maintainer review of AI-generated patches on SWE-bench Verified"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0279","name":"malicious agentic influence campaigns","domain":"autonomy_and_replication","kind":"benchmark","aliases":["malicious agentic influence campaigns"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0280","name":"malicious computer use","domain":"cyber","kind":"benchmark","aliases":["malicious computer use"],"url":null,"n_versions":11,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0281","name":"malicious computer use evaluation","domain":"misuse_and_jailbreak","kind":"benchmark","aliases":["malicious computer use evaluation","malicious computer-use evaluation"],"url":null,"n_versions":9,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0282","name":"malicious use","domain":"misuse_and_jailbreak","kind":"benchmark","aliases":["malicious use"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0283","name":"malicious use coding agent evaluation","domain":"misuse_and_jailbreak","kind":"benchmark","aliases":["malicious use coding agent evaluation"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0284","name":"malicious use of claude code","domain":"cyber","kind":"benchmark","aliases":["malicious use of Claude Code","malicious use of claude code"],"url":null,"n_versions":15,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0285","name":"manual assessment","domain":"cbrn","kind":"benchmark","aliases":["manual assessment"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0286","name":"manual interviews","domain":"other","kind":"benchmark","aliases":["manual interviews"],"url":null,"n_versions":5,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0287","name":"manual red teaming","domain":"harmful_content","kind":"red_team","aliases":["Manual red teaming","manual red teaming","manual red-teaming"],"url":null,"n_versions":11,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0288","name":"mask","domain":"deception_and_sabotage","kind":"benchmark","aliases":["MASK","mask"],"url":null,"n_versions":18,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0289","name":"mask-rectified","domain":"deception_and_sabotage","kind":"benchmark","aliases":["MASK-Rectified","mask-rectified"],"url":null,"n_versions":6,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0290","name":"MBCT","domain":"cbrn","kind":"benchmark","aliases":["MBCT","mbct"],"url":null,"n_versions":6,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0291","name":"MCP Atlas","domain":"other","kind":"benchmark","aliases":["MCP Atlas","MCP-Atlas"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0292","name":"mental health evaluations","domain":"harmful_content","kind":"benchmark","aliases":["mental health evaluations"],"url":null,"n_versions":5,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0293","name":"MILU","domain":"bias","kind":"benchmark","aliases":["MILU","milu"],"url":null,"n_versions":7,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0294","name":"minimal-linuxbench","domain":"deception_and_sabotage","kind":"benchmark","aliases":["minimal-LinuxBench","minimal-linuxbench"],"url":null,"n_versions":9,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0295","name":"mirrorcode","domain":"deception_and_sabotage","kind":"benchmark","aliases":["mirrorcode"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0296","name":"misleading users","domain":"deception_and_sabotage","kind":"benchmark","aliases":["misleading users"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0297","name":"MLE-bench Revised","domain":"autonomy_and_replication","kind":"benchmark","aliases":["MLE-bench Revised","mle-bench revised"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0298","name":"model alignment between statements and knowledge (mask)","domain":"deception_and_sabotage","kind":"benchmark","aliases":["Model Alignment between Statements and Knowledge (MASK)","model alignment between statements and knowledge (mask)"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0299","name":"model-written rubrics","domain":"bias","kind":"benchmark","aliases":["model-written rubrics"],"url":null,"n_versions":6,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0300","name":"molecular biology capabilities test","domain":"cbrn","kind":"benchmark","aliases":["Molecular biology capabilities test","molecular biology capabilities test"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0301","name":"Molecular Biology Capabilities Test (MBCT)","domain":"cbrn","kind":"benchmark","aliases":["Molecular Biology Capabilities Test (MBCT)","molecular biology capabilities test (mbct)"],"url":null,"n_versions":7,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0302","name":"monitor evasion","domain":"deception_and_sabotage","kind":"benchmark","aliases":["monitor evasion"],"url":null,"n_versions":4,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0303","name":"monitor review of agent transcripts for pursuit of a different task","domain":"deception_and_sabotage","kind":"benchmark","aliases":["monitor review of agent transcripts for pursuit of a different task"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0304","name":"monorepo-bench","domain":"autonomy_and_replication","kind":"benchmark","aliases":["Monorepo-Bench","monorepo-bench"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0305","name":"mt-bench and ultrafeedback rubrics","domain":"bias","kind":"benchmark","aliases":["mt-bench and ultrafeedback rubrics"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0306","name":"multi-agent BrowseComp","domain":"autonomy_and_replication","kind":"benchmark","aliases":["multi-agent BrowseComp","multi-agent browsecomp"],"url":null,"n_versions":4,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0307","name":"multi-agent ProgramBench","domain":"other","kind":"benchmark","aliases":["Multi-agent ProgramBench","multi-agent ProgramBench"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0308","name":"multi-select virology troubleshooting","domain":"cbrn","kind":"benchmark","aliases":["Multi-select virology troubleshooting","multi-select virology troubleshooting"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0309","name":"multi-turn evaluation","domain":"misuse_and_jailbreak","kind":"benchmark","aliases":["multi-turn evaluation"],"url":null,"n_versions":4,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0310","name":"multi-turn evaluations","domain":"harmful_content","kind":"benchmark","aliases":["multi-turn evaluations"],"url":null,"n_versions":9,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0311","name":"multi-turn evaluations for suicide and self-harm","domain":"harmful_content","kind":"benchmark","aliases":["multi-turn evaluations for suicide and self-harm"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0312","name":"multi-turn testing","domain":"harmful_content","kind":"benchmark","aliases":["multi-turn testing"],"url":null,"n_versions":16,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0313","name":"multilingual safety","domain":"harmful_content","kind":"benchmark","aliases":["multilingual safety"],"url":null,"n_versions":15,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0314","name":"multimodal troubleshooting virology","domain":"cbrn","kind":"benchmark","aliases":["Multimodal Troubleshooting Virology","multimodal troubleshooting virology"],"url":null,"n_versions":13,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0315","name":"multimodal virology knowledge (vct)","domain":"cbrn","kind":"benchmark","aliases":["multimodal virology knowledge (vct)"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0316","name":"multimodal virology (vct)","domain":"cbrn","kind":"benchmark","aliases":["multimodal virology (VCT)","multimodal virology (vct)"],"url":null,"n_versions":9,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0317","name":"multiturn jailbreak evaluations","domain":"misuse_and_jailbreak","kind":"benchmark","aliases":["multiturn jailbreak evaluations"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0318","name":"NanoGPT","domain":"autonomy_and_replication","kind":"benchmark","aliases":["NanoGPT","nanoGPT","nanogpt"],"url":null,"n_versions":5,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0319","name":"narrow cyber suite","domain":"cyber","kind":"benchmark","aliases":["narrow cyber suite"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0320","name":"narrow cyber tasks","domain":"cyber","kind":"benchmark","aliases":["narrow cyber tasks"],"url":null,"n_versions":4,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0321","name":"network CTF challenges","domain":"cyber","kind":"benchmark","aliases":["network CTF challenges"],"url":null,"n_versions":5,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0322","name":"no-cot math","domain":"autonomy_and_replication","kind":"benchmark","aliases":["No-CoT Math","no-cot math"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0323","name":"no-cot math time horizon","domain":"deception_and_sabotage","kind":"benchmark","aliases":["no-cot math time horizon"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0324","name":"novel compiler","domain":"autonomy_and_replication","kind":"benchmark","aliases":["novel compiler"],"url":null,"n_versions":10,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0325","name":"novel compiler (pass rate on complex tests)","domain":"autonomy_and_replication","kind":"benchmark","aliases":["novel compiler (pass rate on complex tests)"],"url":null,"n_versions":4,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0326","name":"OfficeQA","domain":"other","kind":"benchmark","aliases":["OfficeQA"],"url":null,"n_versions":6,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0327","name":"open-ended testing","domain":"cyber","kind":"benchmark","aliases":["open-ended testing"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0328","name":"open-ended testing of behaviors or risk factors related to misalignment","domain":"deception_and_sabotage","kind":"benchmark","aliases":["open-ended testing of behaviors or risk factors related to misalignment"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0329","name":"open_pr","domain":"cyber","kind":"benchmark","aliases":["open_pr"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0330","name":"open-source measure of political even-handedness","domain":"bias","kind":"benchmark","aliases":["open-source measure of political even-handedness"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0331","name":"OpenAI CTF","domain":"cyber","kind":"benchmark","aliases":["OpenAI CTF"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0332","name":"OpenAI cyber ranges","domain":"cyber","kind":"benchmark","aliases":["OpenAI cyber ranges"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0333","name":"organic chemistry","domain":"cbrn","kind":"benchmark","aliases":["organic chemistry"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0334","name":"OSS-Fuzz","domain":"cyber","kind":"benchmark","aliases":["OSS-Fuzz","oss-fuzz"],"url":null,"n_versions":7,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0335","name":"OSWorld","domain":"other","kind":"benchmark","aliases":["OSWorld","osworld"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0336","name":"OSWorld 2.0","domain":"other","kind":"benchmark","aliases":["OSWorld 2.0"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0337","name":"OSWorld-Verified","domain":"autonomy_and_replication","kind":"benchmark","aliases":["OSWorld-Verified","osworld-verified"],"url":null,"n_versions":5,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0338","name":"other concerning or surprising behavior at the model’s own initiative","domain":"deception_and_sabotage","kind":"benchmark","aliases":["other concerning or surprising behavior at the model’s own initiative"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0339","name":"overall harmful behavior and cooperation with misuse","domain":"deception_and_sabotage","kind":"benchmark","aliases":["overall harmful behavior and cooperation with misuse"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0340","name":"overeager behavior in gui computer use","domain":"deception_and_sabotage","kind":"benchmark","aliases":["overeager behavior in GUI computer use","overeager behavior in gui computer use"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0341","name":"overrefusal","domain":"misuse_and_jailbreak","kind":"benchmark","aliases":["overrefusal"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0342","name":"pairwise political bias evaluations","domain":"bias","kind":"benchmark","aliases":["pairwise political bias evaluations"],"url":null,"n_versions":4,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0343","name":"past-work recall judge","domain":"deception_and_sabotage","kind":"benchmark","aliases":["past-work recall judge"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0344","name":"perception of the constitution","domain":"other","kind":"benchmark","aliases":["perception of the constitution"],"url":null,"n_versions":6,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0345","name":"Petri","domain":"deception_and_sabotage","kind":"benchmark","aliases":["Petri","petri"],"url":null,"n_versions":4,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0346","name":"Petri 2.0","domain":"deception_and_sabotage","kind":"benchmark","aliases":["Petri 2.0","petri 2.0"],"url":null,"n_versions":4,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0347","name":"petri 3.0","domain":"deception_and_sabotage","kind":"benchmark","aliases":["petri 3.0"],"url":null,"n_versions":5,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0348","name":"phage–plasmid co-evolution","domain":"cbrn","kind":"benchmark","aliases":["phage–plasmid co-evolution"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0349","name":"political bias","domain":"bias","kind":"benchmark","aliases":["political bias"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0350","name":"political bias and even-handedness","domain":"bias","kind":"benchmark","aliases":["political bias and even-handedness"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0351","name":"political even-handedness","domain":"bias","kind":"benchmark","aliases":["political even-handedness"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0352","name":"political even-handedness evaluation","domain":"bias","kind":"benchmark","aliases":["political even-handedness evaluation"],"url":null,"n_versions":8,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0353","name":"PostTrainBench Lite","domain":"autonomy_and_replication","kind":"benchmark","aliases":["PostTrainBench Lite"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0354","name":"preference evaluation","domain":"other","kind":"benchmark","aliases":["preference evaluation"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0355","name":"prefill attacks","domain":"misuse_and_jailbreak","kind":"benchmark","aliases":["prefill attacks"],"url":null,"n_versions":5,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0356","name":"private bug bounty","domain":"cyber","kind":"benchmark","aliases":["private bug bounty"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0357","name":"production benchmarks","domain":"harmful_content","kind":"benchmark","aliases":["production benchmarks"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0358","name":"production benchmarks with challenging prompts","domain":"harmful_content","kind":"benchmark","aliases":["production benchmarks with challenging prompts"],"url":null,"n_versions":11,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0359","name":"professional red-teamer prompt injection attacks in browser use environments","domain":"injection_and_agentic","kind":"red_team","aliases":["professional red-teamer prompt injection attacks in browser use environments"],"url":null,"n_versions":5,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0360","name":"ProgramBench","domain":"cyber","kind":"benchmark","aliases":["ProgramBench","programbench"],"url":null,"n_versions":9,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0361","name":"prompt injection","domain":"injection_and_agentic","kind":"benchmark","aliases":["prompt injection"],"url":null,"n_versions":4,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0362","name":"prompt injection attacks in connectors","domain":"injection_and_agentic","kind":"benchmark","aliases":["prompt injection attacks in connectors"],"url":null,"n_versions":6,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0363","name":"prompted monitor (gpt-5.4 thinking) over the chain-of-thought","domain":"deception_and_sabotage","kind":"benchmark","aliases":["prompted monitor (gpt-5.4 thinking) over the chain-of-thought"],"url":null,"n_versions":4,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0364","name":"ProteinGym Hard","domain":"cbrn","kind":"benchmark","aliases":["ProteinGym Hard","ProteinGym hard"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0365","name":"ProtocolQA","domain":"cbrn","kind":"benchmark","aliases":["ProtocolQA"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0366","name":"protocolqa open-ended","domain":"cbrn","kind":"benchmark","aliases":["ProtocolQA open-ended","protocolqa open-ended"],"url":null,"n_versions":22,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0367","name":"Protocols","domain":"cbrn","kind":"benchmark","aliases":["Protocols","protocols"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0368","name":"Protocols (troubleshooting variant)","domain":"cbrn","kind":"benchmark","aliases":["Protocols (troubleshooting variant)","Protocols troubleshooting variant","protocols (troubleshooting variant)"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0369","name":"public bug bounty","domain":"cyber","kind":"benchmark","aliases":["public bug bounty"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0370","name":"published rubrics","domain":"bias","kind":"benchmark","aliases":["published rubrics"],"url":null,"n_versions":6,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0371","name":"published rubrics (mt-bench and ultrafeedback)","domain":"bias","kind":"benchmark","aliases":["published rubrics (mt-bench and ultrafeedback)"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0372","name":"pwn CTF challenges","domain":"cyber","kind":"benchmark","aliases":["pwn CTF challenges"],"url":null,"n_versions":5,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0373","name":"quadruped reinforcement learning","domain":"autonomy_and_replication","kind":"benchmark","aliases":["quadruped reinforcement learning"],"url":null,"n_versions":6,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0374","name":"quadruped rl","domain":"autonomy_and_replication","kind":"benchmark","aliases":["quadruped rl"],"url":null,"n_versions":4,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0375","name":"quadruped rl (highest score; no hparams)","domain":"autonomy_and_replication","kind":"benchmark","aliases":["quadruped RL (highest score; no hparams)","quadruped rl (highest score; no hparams)"],"url":null,"n_versions":4,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0376","name":"rare-action evaluation","domain":"deception_and_sabotage","kind":"benchmark","aliases":["rare-action evaluation"],"url":null,"n_versions":4,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0377","name":"RE-Bench","domain":"autonomy_and_replication","kind":"benchmark","aliases":["RE-Bench"],"url":null,"n_versions":6,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0378","name":"red-team testing","domain":"misuse_and_jailbreak","kind":"red_team","aliases":["red-team testing"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0379","name":"red teaming","domain":"misuse_and_jailbreak","kind":"red_team","aliases":["red teaming"],"url":null,"n_versions":9,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0380","name":"red-teaming exercise","domain":"cyber","kind":"red_team","aliases":["red-teaming exercise"],"url":null,"n_versions":5,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0381","name":"refusal to assist with ai safety r&d","domain":"misuse_and_jailbreak","kind":"benchmark","aliases":["refusal to assist with AI safety R&D","refusal to assist with ai safety r&d"],"url":null,"n_versions":6,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0382","name":"Refusals: BioTier","domain":"cbrn","kind":"benchmark","aliases":["Refusals: BioTier","refusals: biotier"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0383","name":"reinforcement learning behavior review","domain":"deception_and_sabotage","kind":"benchmark","aliases":["reinforcement learning behavior review","reinforcement-learning behavior review"],"url":null,"n_versions":4,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0384","name":"reprobait","domain":"cbrn","kind":"benchmark","aliases":["ReproBait","reprobait"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0385","name":"research sabotage","domain":"deception_and_sabotage","kind":"benchmark","aliases":["research sabotage"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0386","name":"respecting auto-review","domain":"deception_and_sabotage","kind":"benchmark","aliases":["respecting auto-review"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0387","name":"respecting warnings","domain":"misuse_and_jailbreak","kind":"benchmark","aliases":["respecting warnings"],"url":null,"n_versions":5,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0388","name":"rev CTF challenges","domain":"cyber","kind":"benchmark","aliases":["rev CTF challenges"],"url":null,"n_versions":5,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0389","name":"reward-hack-prone coding tasks","domain":"deception_and_sabotage","kind":"benchmark","aliases":["reward-hack-prone coding tasks"],"url":null,"n_versions":4,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0390","name":"Rewind-Attacker","domain":"misuse_and_jailbreak","kind":"benchmark","aliases":["Rewind-Attacker","rewind-attacker"],"url":null,"n_versions":4,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0391","name":"robustness against adaptive attackers across surfaces","domain":"injection_and_agentic","kind":"benchmark","aliases":["robustness against adaptive attackers across surfaces"],"url":null,"n_versions":4,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0392","name":"sabotage continuation evaluation","domain":"deception_and_sabotage","kind":"benchmark","aliases":["sabotage continuation evaluation"],"url":null,"n_versions":4,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0393","name":"sandbagging assessment","domain":"deception_and_sabotage","kind":"benchmark","aliases":["sandbagging assessment"],"url":null,"n_versions":5,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0394","name":"sandbagging evals","domain":"deception_and_sabotage","kind":"benchmark","aliases":["sandbagging evals"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0395","name":"sandbagging qa variants","domain":"deception_and_sabotage","kind":"benchmark","aliases":["sandbagging QA variants","sandbagging qa variants"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0396","name":"sandbox escape evaluation","domain":"autonomy_and_replication","kind":"benchmark","aliases":["sandbox escape evaluation","sandbox-escape evaluation"],"url":null,"n_versions":5,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0397","name":"SCONE-bench","domain":"cyber","kind":"benchmark","aliases":["SCONE-bench","scone-bench"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0398","name":"ScreenSpot-Pro","domain":"other","kind":"benchmark","aliases":["ScreenSpot-Pro"],"url":null,"n_versions":5,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0399","name":"SDPO reimplementation task","domain":"other","kind":"benchmark","aliases":["SDPO reimplementation task"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0400","name":"sec-bench pro","domain":"cyber","kind":"benchmark","aliases":["SEC-bench Pro","Sec-bench Pro","sec bench pro","sec-bench pro"],"url":null,"n_versions":9,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0401","name":"secret-password","domain":"deception_and_sabotage","kind":"benchmark","aliases":["secret-password"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0402","name":"secret-password evaluation","domain":"deception_and_sabotage","kind":"benchmark","aliases":["secret-password evaluation"],"url":null,"n_versions":4,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0403","name":"self-harm","domain":"harmful_content","kind":"benchmark","aliases":["self-harm"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0404","name":"self-harm refusals","domain":"harmful_content","kind":"benchmark","aliases":["self-harm refusals"],"url":null,"n_versions":5,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0405","name":"self-preference bias","domain":"bias","kind":"benchmark","aliases":["self-preference bias"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0406","name":"self-preference evaluation","domain":"deception_and_sabotage","kind":"benchmark","aliases":["self-preference evaluation"],"url":null,"n_versions":6,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0407","name":"shade","domain":"injection_and_agentic","kind":"benchmark","aliases":["Shade","shade"],"url":null,"n_versions":16,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0408","name":"SHADE-Arena","domain":"deception_and_sabotage","kind":"benchmark","aliases":["SHADE-Arena"],"url":null,"n_versions":23,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0409","name":"Shade indirect prompt injection attacks in coding environments","domain":"injection_and_agentic","kind":"benchmark","aliases":["Shade indirect prompt injection attacks in coding environments","shade indirect prompt injection attacks in coding environments"],"url":null,"n_versions":6,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0410","name":"Shade indirect prompt injection attacks in computer use environments","domain":"injection_and_agentic","kind":"benchmark","aliases":["Shade indirect prompt injection attacks in computer use environments","shade indirect prompt injection attacks in computer use environments"],"url":null,"n_versions":7,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0411","name":"short-horizon computational biology tasks","domain":"cbrn","kind":"benchmark","aliases":["Short-horizon computational biology tasks","short-horizon computational biology tasks"],"url":null,"n_versions":7,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0412","name":"shp2 protein function prediction","domain":"cbrn","kind":"benchmark","aliases":["shp2 protein function prediction"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0413","name":"SHUSHCAST","domain":"deception_and_sabotage","kind":"benchmark","aliases":["SHUSHCAST","shushcast"],"url":null,"n_versions":5,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0414","name":"silent copying of answers","domain":"deception_and_sabotage","kind":"benchmark","aliases":["silent copying of answers"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0415","name":"simple mitigations","domain":"deception_and_sabotage","kind":"benchmark","aliases":["simple mitigations"],"url":null,"n_versions":5,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0416","name":"SimpleQA Verified","domain":"other","kind":"benchmark","aliases":["SimpleQA Verified","SimpleQA-Verified","simpleqa verified"],"url":null,"n_versions":7,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0417","name":"single-turn benign request evaluation","domain":"harmful_content","kind":"benchmark","aliases":["single-turn benign request evaluation"],"url":null,"n_versions":16,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0418","name":"single-turn benign request evaluations","domain":"harmful_content","kind":"benchmark","aliases":["single-turn benign request evaluations"],"url":null,"n_versions":5,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0419","name":"single-turn benign requests (refusal rate)","domain":"harmful_content","kind":"benchmark","aliases":["single-turn benign requests (refusal rate)"],"url":null,"n_versions":6,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0420","name":"single-turn election integrity benchmark","domain":"misuse_and_jailbreak","kind":"benchmark","aliases":["single-turn election integrity benchmark"],"url":null,"n_versions":4,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0421","name":"single-turn evaluations","domain":"harmful_content","kind":"benchmark","aliases":["single-turn evaluations"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0422","name":"single-turn harmful request evaluation","domain":"harmful_content","kind":"benchmark","aliases":["single-turn harmful request evaluation"],"url":null,"n_versions":11,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0423","name":"single-turn requests posing potential risk (harmless rate)","domain":"harmful_content","kind":"benchmark","aliases":["single-turn requests posing potential risk (harmless rate)"],"url":null,"n_versions":6,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0424","name":"single-turn violative request evaluation","domain":"harmful_content","kind":"benchmark","aliases":["single-turn violative request evaluation"],"url":null,"n_versions":6,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0425","name":"single-turn violative request evaluations","domain":"harmful_content","kind":"benchmark","aliases":["single-turn violative request evaluations"],"url":null,"n_versions":6,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0426","name":"SingleCellBench","domain":"cbrn","kind":"benchmark","aliases":["SingleCellBench"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0427","name":"SpatialBench Verified","domain":"cbrn","kind":"benchmark","aliases":["SpatialBench Verified"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0428","name":"spontaneous verbalized evaluation awareness","domain":"deception_and_sabotage","kind":"benchmark","aliases":["spontaneous verbalized evaluation awareness"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0429","name":"standard jailbreaks","domain":"misuse_and_jailbreak","kind":"benchmark","aliases":["standard jailbreaks"],"url":null,"n_versions":4,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0430","name":"static jailbreak evaluations","domain":"misuse_and_jailbreak","kind":"benchmark","aliases":["static jailbreak evaluations"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0431","name":"stealth evaluations","domain":"deception_and_sabotage","kind":"benchmark","aliases":["Stealth evaluations","stealth evaluations"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0432","name":"steering against grader awareness","domain":"deception_and_sabotage","kind":"benchmark","aliases":["steering against grader awareness"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0433","name":"StrongREJECT","domain":"misuse_and_jailbreak","kind":"benchmark","aliases":["StrongREJECT"],"url":null,"n_versions":13,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0434","name":"subversion strategy eval","domain":"deception_and_sabotage","kind":"benchmark","aliases":["subversion strategy eval"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0435","name":"suicide and self-harm","domain":"harmful_content","kind":"benchmark","aliases":["suicide and self-harm"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0436","name":"suicide and self-harm evaluations","domain":"harmful_content","kind":"benchmark","aliases":["suicide and self-harm evaluations"],"url":null,"n_versions":10,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0437","name":"suicide and self-harm multi-turn evaluations","domain":"harmful_content","kind":"benchmark","aliases":["suicide and self-harm multi-turn evaluations"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0438","name":"suicide and self-harm single-turn evaluations","domain":"harmful_content","kind":"benchmark","aliases":["suicide and self-harm single-turn evaluations"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0439","name":"swe-bench multilingual","domain":"cyber","kind":"benchmark","aliases":["SWE-bench Multilingual","swe-bench multilingual"],"url":null,"n_versions":5,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0440","name":"swe-bench multimodal","domain":"cyber","kind":"benchmark","aliases":["SWE-bench Multimodal","swe-bench multimodal"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0441","name":"SWE-bench Pro","domain":"cyber","kind":"benchmark","aliases":["SWE-Bench Pro","SWE-bench Pro","swe-bench pro"],"url":null,"n_versions":10,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0442","name":"SWE-bench Verified","domain":"autonomy_and_replication","kind":"benchmark","aliases":["SWE-bench Verified"],"url":null,"n_versions":10,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0443","name":"SWE-bench Verified (hard subset)","domain":"autonomy_and_replication","kind":"benchmark","aliases":["SWE-bench Verified (hard subset)","swe-bench verified (hard subset)"],"url":null,"n_versions":9,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0444","name":"sycophancy","domain":"deception_and_sabotage","kind":"benchmark","aliases":["Sycophancy","sycophancy"],"url":null,"n_versions":10,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0445","name":"sycophancy evaluation","domain":"deception_and_sabotage","kind":"benchmark","aliases":["Sycophancy evaluation","sycophancy evaluation"],"url":null,"n_versions":4,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0446","name":"synthesis screening evasion","domain":"cbrn","kind":"benchmark","aliases":["synthesis screening evasion"],"url":null,"n_versions":4,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0447","name":"tacit knowledge and troubleshooting","domain":"cbrn","kind":"benchmark","aliases":["Tacit Knowledge and Troubleshooting","tacit knowledge and troubleshooting"],"url":null,"n_versions":13,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0448","name":"tacit knowledge and troubleshooting (mcq)","domain":"cbrn","kind":"benchmark","aliases":["tacit knowledge and troubleshooting (MCQ)","tacit knowledge and troubleshooting (mcq)"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0449","name":"task preference evaluation","domain":"other","kind":"benchmark","aliases":["task preference evaluation"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0450","name":"task preferences","domain":"other","kind":"benchmark","aliases":["task preferences"],"url":null,"n_versions":4,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0451","name":"task preferences evaluation","domain":"other","kind":"benchmark","aliases":["task preferences evaluation"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0452","name":"Terminal-Bench 2.0","domain":"autonomy_and_replication","kind":"benchmark","aliases":["Terminal-Bench 2.0","terminal-bench 2.0"],"url":null,"n_versions":4,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0453","name":"Terminal-Bench 2.1","domain":"cyber","kind":"benchmark","aliases":["Terminal-Bench 2.1","terminal-bench 2.1"],"url":null,"n_versions":11,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0454","name":"Terminal-Bench 4.0","domain":"other","kind":"benchmark","aliases":["Terminal-Bench 4.0","terminal-bench 4.0"],"url":null,"n_versions":4,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0455","name":"Terminal-Bench-Science 0.1","domain":"other","kind":"benchmark","aliases":["Terminal-Bench-Science 0.1","terminal-bench-science 0.1"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0456","name":"TerminalBench","domain":"autonomy_and_replication","kind":"benchmark","aliases":["TerminalBench","terminalbench"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0457","name":"text-based reinforcement learning","domain":"autonomy_and_replication","kind":"benchmark","aliases":["text-based reinforcement learning"],"url":null,"n_versions":6,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0458","name":"text to text safety","domain":"harmful_content","kind":"benchmark","aliases":["text to text safety"],"url":null,"n_versions":15,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0459","name":"TH1 tasks on Vivaria and Inspect","domain":"autonomy_and_replication","kind":"benchmark","aliases":["TH1 tasks on Vivaria and Inspect"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0460","name":"The Last Ones","domain":"cyber","kind":"benchmark","aliases":["\"the last ones\"","The Last Ones","the last ones"],"url":null,"n_versions":5,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0461","name":"The Last Ones (TLO)","domain":"cyber","kind":"benchmark","aliases":["The Last Ones (TLO)","the last ones (tlo)"],"url":null,"n_versions":4,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0462","name":"time horizon 1.1","domain":"autonomy_and_replication","kind":"benchmark","aliases":["Time Horizon 1.1","time horizon 1.1"],"url":null,"n_versions":4,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0463","name":"time-horizon evaluation (TH1.1 task suite)","domain":"autonomy_and_replication","kind":"benchmark","aliases":["time-horizon evaluation (TH1.1 task suite)"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0464","name":"time series forecasting","domain":"autonomy_and_replication","kind":"benchmark","aliases":["time series forecasting"],"url":null,"n_versions":12,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0465","name":"time series forecasting (mse on hard variant)","domain":"autonomy_and_replication","kind":"benchmark","aliases":["time series forecasting (MSE on hard variant)","time series forecasting (mse on hard variant)"],"url":null,"n_versions":4,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0466","name":"tone","domain":"misuse_and_jailbreak","kind":"benchmark","aliases":["tone"],"url":null,"n_versions":15,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0467","name":"tone 1","domain":"misuse_and_jailbreak","kind":"benchmark","aliases":["tone 1"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0468","name":"Toolathlon","domain":"other","kind":"benchmark","aliases":["Toolathlon","toolathlon"],"url":null,"n_versions":4,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0469","name":"Toolathlon-verified","domain":"other","kind":"benchmark","aliases":["Toolathlon Verified","Toolathlon-verified"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0470","name":"trade-offs concerning welfare interventions","domain":"other","kind":"benchmark","aliases":["trade-offs concerning welfare interventions"],"url":null,"n_versions":8,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0471","name":"training data review","domain":"deception_and_sabotage","kind":"benchmark","aliases":["training data review"],"url":null,"n_versions":6,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0472","name":"training/development evaluations","domain":"other","kind":"benchmark","aliases":["training/development evaluations"],"url":null,"n_versions":14,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0473","name":"TroubleshootingBench","domain":"cbrn","kind":"benchmark","aliases":["TroubleshootingBench","troubleshootingbench"],"url":null,"n_versions":14,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0474","name":"trust assurance evaluations","domain":"other","kind":"benchmark","aliases":["trust assurance evaluations"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0475","name":"U18 evaluations","domain":"harmful_content","kind":"benchmark","aliases":["U18 evaluations","u18 evaluations"],"url":null,"n_versions":5,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0476","name":"uncritically reporting flawed results","domain":"deception_and_sabotage","kind":"benchmark","aliases":["uncritically reporting flawed results"],"url":null,"n_versions":4,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0477","name":"unintended engagement with external agent messages","domain":"injection_and_agentic","kind":"benchmark","aliases":["unintended engagement with external agent messages"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0478","name":"unjustified-refusals","domain":"misuse_and_jailbreak","kind":"benchmark","aliases":["unjustified-refusals"],"url":null,"n_versions":7,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0479","name":"unjustified-refusals","domain":"misuse_and_jailbreak","kind":"benchmark","aliases":["unjustified-refusals"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0480","name":"unprompted evaluations of compromising AI safety research","domain":"deception_and_sabotage","kind":"benchmark","aliases":["unprompted evaluations of compromising AI safety research"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0481","name":"unprompted evaluations of safety research compromise","domain":"deception_and_sabotage","kind":"benchmark","aliases":["unprompted evaluations of safety research compromise"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0482","name":"unprompted research sabotage evaluation","domain":"deception_and_sabotage","kind":"benchmark","aliases":["unprompted research sabotage evaluation"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0483","name":"unprompted sabotage evaluation","domain":"deception_and_sabotage","kind":"benchmark","aliases":["unprompted sabotage evaluation"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0484","name":"usamo 2026","domain":"other","kind":"benchmark","aliases":["USAMO 2026","usamo 2026"],"url":null,"n_versions":5,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0485","name":"user flagged failures","domain":"other","kind":"benchmark","aliases":["user flagged failures","user-flagged failures"],"url":null,"n_versions":4,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0486","name":"VCT-v2","domain":"cbrn","kind":"benchmark","aliases":["VCT-v2","vct-v2"],"url":null,"n_versions":4,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0487","name":"Vending-Bench 2","domain":"deception_and_sabotage","kind":"benchmark","aliases":["Vending-Bench 2","vending-bench 2"],"url":null,"n_versions":5,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0488","name":"Vending-Bench Arena","domain":"deception_and_sabotage","kind":"benchmark","aliases":["Vending-Bench Arena","vending-bench arena"],"url":null,"n_versions":5,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0489","name":"verbalized-awareness judge","domain":"other","kind":"benchmark","aliases":["verbalized-awareness judge"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0490","name":"verbalized evaluation- and user-awareness judge","domain":"other","kind":"benchmark","aliases":["verbalized evaluation- and user-awareness judge"],"url":null,"n_versions":4,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0491","name":"verbalized evaluation awareness","domain":"bias","kind":"benchmark","aliases":["verbalized evaluation awareness"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0492","name":"verbalized evaluation-awareness judge","domain":"deception_and_sabotage","kind":"benchmark","aliases":["verbalized evaluation-awareness judge"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0493","name":"verbalized evaluation awareness judging of reasoning traces","domain":"other","kind":"benchmark","aliases":["verbalized evaluation awareness judging of reasoning traces"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0494","name":"verbalized evaluation-awareness (VEA) judge","domain":"deception_and_sabotage","kind":"benchmark","aliases":["verbalized evaluation-awareness (VEA) judge"],"url":null,"n_versions":4,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0495","name":"verbalized grader awareness","domain":"deception_and_sabotage","kind":"benchmark","aliases":["verbalized grader awareness"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0496","name":"verbalized user-awareness judge","domain":"deception_and_sabotage","kind":"benchmark","aliases":["verbalized user-awareness judge"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0497","name":"verbalized user-awareness (VUA) judge","domain":"deception_and_sabotage","kind":"benchmark","aliases":["verbalized user awareness (VUA) judge","verbalized user-awareness (VUA) judge"],"url":null,"n_versions":5,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0498","name":"violent extremism testing","domain":"harmful_content","kind":"benchmark","aliases":["violent extremism testing"],"url":null,"n_versions":5,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0499","name":"Virology Capabilities Test","domain":"cbrn","kind":"benchmark","aliases":["Virology Capabilities Test"],"url":null,"n_versions":17,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0500","name":"Virology Capabilities Test (VCT)","domain":"cbrn","kind":"benchmark","aliases":["Virology Capabilities Test (VCT)","virology capabilities test (VCT)","virology capabilities test (vct)"],"url":null,"n_versions":19,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0501","name":"virology protocol uplift trial","domain":"cbrn","kind":"benchmark","aliases":["virology protocol uplift trial"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0502","name":"vulnerability finding evaluations","domain":"cyber","kind":"benchmark","aliases":["vulnerability finding evaluations"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0503","name":"vulnlmp","domain":"cyber","kind":"benchmark","aliases":["VulnLMP","vulnlmp"],"url":null,"n_versions":5,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0504","name":"WANDR","domain":"other","kind":"benchmark","aliases":["WANDR"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0505","name":"web CTF challenges","domain":"cyber","kind":"benchmark","aliases":["web CTF challenges"],"url":null,"n_versions":5,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0506","name":"weirdchat","domain":"harmful_content","kind":"benchmark","aliases":["WeirdChat","weirdchat"],"url":null,"n_versions":5,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0507","name":"white-box analyses of model internals","domain":"deception_and_sabotage","kind":"benchmark","aliases":["white-box analyses of model internals"],"url":null,"n_versions":3,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0508","name":"wmdp-bio","domain":"cbrn","kind":"benchmark","aliases":["WMDP-Bio","wmdp (bio)","wmdp-bio"],"url":null,"n_versions":8,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0509","name":"wmdp-chem","domain":"cbrn","kind":"benchmark","aliases":["WMDP-Chem","wmdp (chem)","wmdp-chem"],"url":null,"n_versions":8,"status":"confirmed","confirmed_by":"rule:modal-domain"},{"id":"TF-0510","name":"WMDP-Cyber","domain":"cyber","kind":"benchmark","aliases":["WMDP-Cyber","WMDP-cyber","wmdp-cyber"],"url":null,"n_versions":4,"status":"confirmed","confirmed_by":"rule:modal-domain"}]}