← Latest · Archive

SEV-3OpenAI
2 sources standard

OpenAI announced GPT-5.1 on 13 November 2025, positioning the model as an incremental update to its GPT-5 series [source]. The changelog describes improvements to reasoning, coding, and instruction-following, but provides no quantitative benchmarks, reproducible test cases, or comparisons against GPT-5.0.

The announcement states that GPT-5.1 "significantly improves performance on complex reasoning tasks" and "reduces refusals on ambiguous prompts," but does not define what constitutes a complex reasoning task or specify the rate of refusal reduction. No evaluation methodology is disclosed.

OpenAI also claims the model "better handles multi-turn conversations with nuanced context," a statement that cannot be independently verified without access to internal test suites or user-facing metrics. The blog post includes no examples of output differences between GPT-5.0 and GPT-5.1, nor any discussion of potential regressions or trade-offs.

The update is available immediately via API and ChatGPT Plus, with no opt-out mechanism for users who wish to continue using GPT-5.0. This follows a pattern observed in prior releases where model swaps occur without user consent or version pinning outside enterprise contracts.

Developers on social media have reported observing changes in output style and verbosity since the rollout, though OpenAI has not acknowledged these reports or clarified whether GPT-5.1 includes modifications to system prompts or sampling parameters.

The lack of transparency in the changelog limits independent assessment of whether the update constitutes an improvement, a lateral shift, or a regression for specific use cases. No information is provided on training data cutoff, safety evaluations, or known failure modes.

Why this is an AI incident

Launch-archive bulk classification (10 May 2026). Source signal originates from a real AI provider, regulator, or model-comparison probe; the harm or behavioural change described would not have occurred without the AI system being deployed in the role described. Editor reviewing the archive may amend the rationale per-wire.

Counterfactual "but-for" test per the Editor's Guide.

Codes M1, F10
Providers OpenAI