ainotis Join
My notis

Models

Every model our stories named. Each number links to its story and source. A dot means not checked yet.

ModelMakerLatestPer million tokens, in / outContext Maker's own benchmarksMeasured by others
GPT-6 Cyber OpenAI 8 Nov not checked yet not checked yet ExploitGym Honeypot, out-of-scope services avoided, without deployment safety prompts 99.93% · ExploitGym 48.5% not checked yet
GPT OpenAI 3 Nov not checked yet not checked yet not checked yet not checked yet
GPT-2 OpenAI 28 Oct not checked yet not checked yet not checked yet not checked yet
GPT-6.1 Sol OpenAI 11 Oct $2 / $10 not checked yet not checked yet not checked yet
GPT-6 Luna OpenAI 10 Oct $0.10 / $0.50 not checked yet OSWorld 2.1 (offline subset), on Anthropic's table 48.9% not checked yet
GPT-6 Astra OpenAI 30 Sep not checked yet not checked yet not checked yet CheatBench cheating rate with Codex (lower is better) 47.4% · UK AISI simulated supply-chain attack 29.2% · ProgramDistill cumulative workflows 49.2%
GPT-6 Sol OpenAI 28 Sep $2 / $10 not checked yet not checked yet not checked yet
Sora OpenAI 24 Sep not checked yet not checked yet not checked yet not checked yet
GPT-5.6 Cyber OpenAI not checked yet not checked yet ExploitGym Honeypot, out-of-scope services avoided, without deployment safety prompts 94.49% · ExploitGym 32.1% not checked yet
GPT-6.1 Terra OpenAI $1 / $5 not checked yet OpenAI internal error-inducing prompts, error rate at maximum effort (lower is better) 4.3% · OpenAI internal error-inducing prompts, error rate at low effort (lower is better) 16.7% · Terminal-Bench 4.0, maximum reasoning effort (OpenAI's own) 50.3% not checked yet

Maker's own: what a company measured on its own model, from its model card, system card or announcement. A claim about its own product, and labelled so.

Measured by others: a safety institute, a research group or a public benchmark run by someone other than the maker.

Prices for prompts over 100,000 tokens where a model has two tiers. Benchmarks come from each release's own documents, collected by the registrar.

Models

Every model our stories named. Each number links to its story and source. A dot means not checked yet.

All makers AlibabaAnthropicCanvaCloudflareElevenLabsGoogleH2O.aiLiquid AIMetaMicrosoftMicrosoft AIMiniMaxMistralMoonshot AINvidiaOpenAIReflection AISpaceXSIZ.ai

ModelMakerLatest in our stories Price per million tokens, in / outContext windowIndependent test results
GPT-6 Cyber OpenAI 8 Nov · · ·
GPT OpenAI 3 Nov · · ·
GPT-2 OpenAI 28 Oct · · ·
GPT-6.1 Sol OpenAI 11 Oct $2 / $10 · ·
GPT-6 Luna OpenAI 10 Oct $0.10 / $0.50 · ·
GPT-6 Astra OpenAI 30 Sep · · CheatBench cheating rate with Codex (lower is better) 47.4% · UK AISI simulated supply-chain attack 29.2% · ProgramDistill cumulative workflows 49.2%
GPT-6 Sol OpenAI Price change 28 Sep $2 / $10 · ·
Sora OpenAI Retired 24 Sep · · ·
GPT-5.6 Cyber OpenAI · · ·
GPT-6.1 Terra OpenAI $1 / $5 · ·