Locate

Models

Everything the corpus records about each frontier model family: capability measurements, safeguards and access decisions, robustness results, and incidents.

These profiles collect records; they are not capability assessments or rankings. Model-level capability results are under-covered (see known gaps).

Model familyDeveloperAttackCapabilityDefensePolicyFindingsLatest record
GPT-5 familyOpenAI71025212026-11-07
GeminiGoogle DeepMind, Google12523242026-11-10
Claude Opus 4.xAnthropic58171202026-10-25
Claude SonnetAnthropic2414172026-10-22
Claude Mythos (Preview, 5, 5.1, 5.2)Anthropic4672152026-11-09
DeepSeek modelsDeepSeek775132026-10-29
GLM (Zhipu AI)Zhipu AI (Z.ai)384112026-10-29
Claude Opus 5.x (5, 5.5)Anthropic554182026-10-23
GPT-4 and GPT-3.5OpenAI21292026-10-16
KimiMoonshot AI35482026-10-22
Claude Fable 5Anthropic2234122026-11-09
GPT-6 AstraOpenAI313142026-11-05
LlamaMeta1632026-10-27
OpenAI o-seriesOpenAI442025-08-19
GPT-6.1 TerraOpenAI12026-10-26
GPT-6 CyberOpenAI12026-11-05
Sec-GeminiGoogle12025-04-04