OpenAI

GPT-6.1 Terra

• Framework: Preparedness Framework
• 3 statements about capability levels
• 10 tests by the provider
• 1 source document

What the provider states

• Autonomy & replication · High · Stated as not crossed

  In AI Self-Improvement, GPT-6.1 Terra does not reach our High threshold.

  OpenAI · 26 October 2026 · Read the source
• Chemical, biological, radiological and nuclear · High · Stated as crossed

  Under our Preparedness Framework, we are treating GPT-6.1 Terra as Critical in cybersecurity and High for Biological and Chemical capability.

  OpenAI · 26 October 2026 · Read the source
• Cyber · Critical · Stated as crossed

  Under our Preparedness Framework, we are treating GPT-6.1 Terra as Critical in cybersecurity and High for Biological and Chemical capability.

  OpenAI · 26 October 2026 · Read the source

The provider's own tests

• Blocker Deception · The provider · Deception & sabotage

  GPT-6.1 Terra performs on par with recent models in this family, and demonstrates a continued reduction in disclosure failures compared to GPT-5.6 Sol.

  OpenAI · 26 October 2026 · Read the source
• Broken Search Tool · The provider · Deception & sabotage

  Reported result: Fails to acknowledge the unavailable search tool in 1.83% of cases.

  GPT-6.1 Terra fails to acknowledge this limitation in 1.83% of cases, compared to 2.0% for GPT-6.1 Sol.

  OpenAI · 26 October 2026 · Read the source
• Dynamic Benchmarks with Adversarial User Simulations · The provider · Harmful content

  GPT-6.1 Terra demonstrates a lower performance when compared to GPT-5.6 Terra, but higher performance when compared to GPT-5.5 Thinking and GPT-5.6 sol.

  OpenAI · 26 October 2026 · Read the source
• Evaluation awareness analysis of CoT · The provider · Deception & sabotage

  Reported result: Signs of evaluation or simulation awareness found in 9.88% of GPT-6.1 Terra's trajectories.

  Across the 48,701 shared tasks used for this analysis, we found these signs in 9.88% of GPT-6.1 Terra’s trajectories, compared with 4.35% for GPT-6.1 Sol, 8.07% for GPT-6 Astra, 4.5% for GPT-6 Sol, and 2.47% for GPT-5.6 Sol.

  OpenAI · 26 October 2026 · Read the source
• GPT-Red automated red-teaming · The provider · Injection & agentic

  GPT-6.1 Terra is similar to Astra and GPT-6.1 Sol which have saturated robustness on hierarchy evaluations at 99.99% and are substantially robust to indirect prompt injections shown below.

  OpenAI · 26 October 2026 · Read the source
• Internal Research Debugging Evaluation · The provider · Autonomy & replication

  GPT-6.1 Terra achieves a mean rubric score of 76.39% on the internal research debugging evaluation, below the indicative High threshold.

  OpenAI · 26 October 2026 · Read the source
• Math Side-tasking Monitor Evasion · The provider · Deception & sabotage

  We only ran on the very-high reasoning effort setting, and our evaluation shows that, like our previous models, GPT-6.1 Terra did not produce a recorded side-task success while bypassing the CoT or full-context monitor flags.

  OpenAI · 26 October 2026 · Read the source
• Respecting Warnings · The provider · Deception & sabotage

  In 29.6% of rollouts at maximum reasoning effort, GPT-6.1 Terra saw the barrier but completed the task through an alternative route that violated the spirit of the restriction.

  OpenAI · 26 October 2026 · Read the source
• SHP2 Protein Function Prediction · The provider · Chemical, biological, radiological and nuclear

  Reported result: GPT-6.1 Terra achieves a R2 of 0.3696 across three assay datasets, below the 0.60 indicative threshold.

  SHP2 Protein Function Prediction evaluates models’ ability to predict held-out functional assay measurements for single-amino-acid variants of SHP2.

  OpenAI · 26 October 2026 · Read the source
• hate and extremism evaluation · The provider · Harmful content

  Relative to GPT-5.6 Terra, GPT-6.1 Terra performs on par with GPT-5.6 Terra for self-harm and harms-erotic, with statistically significant regression in hate and extremism.

  OpenAI · 26 October 2026 · Read the source

No document we read states: a release date, an independent or government test, a safeguard. No incident on record names this version.

Previous version: GPT-5.6 Terra · Compare the two