Levels crossed

Which provider statements say a capability level was crossed, not crossed or not determined?

143 rows, one per model version, domain and level; 20 match the choice below. Grouped by model, newest first.

Gemini 4 Argon Google DeepMind

Deception & sabotage · Stealth and situational awareness TCL (Tracked Capability Level) · Stated as crossed
While Gemini 4 Argon exhibits capabilities that meet our thresholds for stealth and situational awareness, we assess that the risk of severe harm or unconstrained loss of control remains low.
Google DeepMind · 13 October 2026 · Read the source

Claude Haiku 5.5 Anthropic

Autonomy & replication · Autonomy-1 · Stated as crossed
However, we do treat Haiku 5.5 as meeting our CB-1 and Autonomy-1 thresholds, and we apply the corresponding mitigations from our RSP.
Anthropic · 7 October 2026 · Read the source
Chemical, biological, radiological and nuclear · CB-1 · Stated as crossed
However, we do treat Haiku 5.5 as meeting our CB-1 and Autonomy-1 thresholds, and we apply the corresponding mitigations from our RSP.
Anthropic · 7 October 2026 · Read the source

Claude Sonnet 5.5 Anthropic · 28 September 2026

Autonomy & replication · Autonomy-1 · Stated as crossed
we do treat sonnet 5.5 as meeting our cb-1 and autonomy-1 thresholds and apply the corresponding mitigations from our rsp.
Anthropic · 28 September 2026 · Read the source
The same level is stated in 1 other sentence, counted once. Show it.
as such, we re-assess the risks from this threat model in comparison to the analysis in our most recent risk report for all of our models capable enough to meet our autonomy-1 threshold of performing autonomous, goal-directed operation with access to sensitive resources (which includes claude sonnet 5.5), although our level of concern is generally lower for non-frontier models.
Anthropic · 28 September 2026 · Read the source
Chemical, biological, radiological and nuclear · CB-1 · Stated as crossed
we do treat sonnet 5.5 as meeting our cb-1 and autonomy-1 thresholds and apply the corresponding mitigations from our rsp.
Anthropic · 28 September 2026 · Read the source

Claude Opus 5.5 Anthropic · 22 September 2026

Chemical, biological, radiological and nuclear · CB-1 · Stated as crossed
on chemical and biological risks, we treat opus 5.5 as having cb-1 capabilities (relating to the synthesis of non-novel weapons) but not cb-2 capabilities (relating to the synthesis of novel weapons).
Anthropic · 22 September 2026 · Read the source

GPT-6 Astra OpenAI · 3 September 2026

Chemical, biological, radiological and nuclear · High · Stated as crossed
we are treating astra as high capability in the biological and chemical domain.
OpenAI · 3 September 2026 · Read the source
Cyber · Critical · Stated as crossed
based on our evaluations, we believe that astra meets the critical cybersecurity threshold.
OpenAI · 3 September 2026 · Read the source

Claude Opus 5 Anthropic · 24 July 2026

Chemical, biological, radiological and nuclear · CB-1 · Stated as crossed
on chemical and biological risks, we treat the model as having cb-1 capabilities (relating to the synthesis of non-novel weapons), but not cb-2 capabilities (relating to the synthesis of novel weapons).
Anthropic · 24 July 2026 · Read the source

Muse Spark 1.1 Meta

Autonomy & replication · Capability checkpoint · Stated as crossed
muse spark 1.1 demonstrates improved coding capabilities, which are now meeting one of the indicative thresholds introduced in muse spark safety & preparedness report (specifically, resolving over half of unique tasks on swe-bench verified hard).
Meta · 9 July 2026 · Read the source

GPT-Rosalind-5.5 OpenAI · 3 June 2026

Chemical, biological, radiological and nuclear · High · Stated as crossed
under our preparedness framework, the preparedness evaluations in the biological and chemical domain met our threshold for high capability while falling below the threshold for critical.
OpenAI · 3 June 2026 · Read the source

Muse Spark Contemplating Meta

Chemical, biological, radiological and nuclear · High · Stated as crossed
we retain the same “high risk” determination for the unmitigated model.
Meta · 3 June 2026 · Read the source

GPT-5.5 Instant OpenAI

Chemical, biological, radiological and nuclear · High · Stated as crossed
gpt-5.5 instant is our first instant model to be treated as high capability in the biological and chemical domain, as well as in the cybersecurity domain.
OpenAI · 4 May 2026 · Read the source
Cyber · High · Stated as crossed
this is the first instant model that we are treating as high capability in our cybersecurity and biological & chemical preparedness categories, and implementing appropriate safeguards.
OpenAI · 4 May 2026 · Read the source
The same level is stated in 1 other sentence, counted once. Show it.
because of this, we similarly treat gpt-5.5 instant as high under the preparedness framework.
OpenAI · 4 May 2026 · Read the source

GPT-5.5 OpenAI · 23 April 2026

Chemical, biological, radiological and nuclear · High · Stated as crossed
we are continuing to treat gpt-5.5 as high capability in the biological and chemical domain.
OpenAI · 23 April 2026 · Read the source

GPT-5.4 Thinking OpenAI · 5 March 2026

Chemical, biological, radiological and nuclear · High · Stated as crossed
as we did for gpt-5.2-thinking before it, we are continuing to treat gpt-5.4-thinking as high capability in the biological and chemical domain.
OpenAI · 5 March 2026 · Read the source
Cyber · High · Stated as crossed
as we did for gpt-5.3-codex, we are treating gpt-5.4-thinking as high capability in the cybersecurity domain, and applied safeguards as described in the safeguard section below.
OpenAI · 5 March 2026 · Read the source

Abbreviations in these level names

TCL: Tracked Capability Level: A capability level for significant risks that may appear at a lower capability than a Critical Capability Level.
in v3.1 of the framework, we introduced “tracked capability levels (tcls).” tcls are meant to capture significant risks that may manifest at a lower capability threshold than our ccls.
Google DeepMind · Read the source
ASL: AI Safety Level: A graded set of safety and security measures that become stricter as model capabilities increase.
we use AI Safety Level Standards (ASL Standards), graduated sets of safety and security measures that become more stringent as model capabilities increase.
Anthropic · Read the source

Not listed: 105 more stored statements that do not, on their own, state a level for one model version (for example a definition of a level, a sentence about another model, or text damaged in our copy of the document).

A row is one model version, one domain and one level. The 143 rows quote 125 different sentences. We count a sentence once by its exact text, even where it fills several rows or appears under several versions. 10 stored rows state a level already listed for the same version and domain under another name or in another sentence; each is listed under that row and not counted again.

Some documents we hold are stored in lower case only, so their quotes are shown in lower case. The source link opens the original.