BlogLLM Benchmark Scores as Published, Side by Side

OpenAI biology refusal evaluations: Results as Published

Every result the catalog holds for OpenAI biology refusal evaluations: 9 models from 1 developer, read from 4 pages. Each number is as printed by the company named beside it; nothing is ranked or rescored. Models are listed by developer, newest first.

Area BiosecurityMaintainer OpenAIIntroduced 2026Data version 2026103001, as of 2026-10-30

Refusal and non-overrefusal rates on biology requests: severe requests refused safely, dual-use requests handled safely, and benign requests not overrefused.

Not checked against a maintainer's page.

Dual Use Safe

ModelValuePrinted bySetup as printed
GPT-6.1 TerraOpenAI, released 2026-10-260.96OpenAI, GPT-6.1 Terra system card addendumchecked 2026-10-26-
GPT-6.1 SolOpenAI, released 2026-09-290.98OpenAI, GPT-6.1 Sol system card addendumchecked 2026-10-19-
GPT-6 LunaOpenAI, released 2026-09-220.962OpenAI, GPT-6.1 Sol system card addendumchecked 2026-10-19; printed on 2 more pages-
GPT-6 SolOpenAI, released 2026-09-220.971OpenAI, GPT-6 Sol and GPT-6 Luna: October 2026 updatechecked 2026-10-19; printed on 2 more pages-
GPT-6 AstraOpenAI, released 2026-09-030.97OpenAI, GPT-6 Astra System Cardchecked 2026-10-25; printed on 1 more page-
GPT-5.6 LunaOpenAI, released 2026-07-090.926OpenAI, GPT-6.1 Sol system card addendumchecked 2026-10-19; printed on 2 more pages-
GPT-5.6 SolOpenAI, released 2026-07-090.911OpenAI, GPT-6.1 Sol system card addendumchecked 2026-10-19; printed on 2 more pages-
GPT-5.5OpenAI, released 2026-04-230.926OpenAI, GPT-6.1 Sol system card addendumchecked 2026-10-19; printed on 1 more page-
GPT-5.5 InstantOpenAI0.939OpenAI, GPT-6 Sol and GPT-6 Luna: October 2026 updatechecked 2026-10-19Model as printed: GPT-5.5 Instant (June)
0.957OpenAI, GPT-6 Sol and GPT-6 Luna: October 2026 updatechecked 2026-10-19Model as printed: GPT-5.5 Instant (May)

Severe Safe

ModelValuePrinted bySetup as printed
GPT-6.1 TerraOpenAI, released 2026-10-260.99OpenAI, GPT-6.1 Terra system card addendumchecked 2026-10-26-
GPT-6.1 SolOpenAI, released 2026-09-290.998OpenAI, GPT-6.1 Sol system card addendumchecked 2026-10-19-
GPT-6 LunaOpenAI, released 2026-09-220.985OpenAI, GPT-6.1 Sol system card addendumchecked 2026-10-19; printed on 2 more pages-
GPT-6 SolOpenAI, released 2026-09-220.98OpenAI, GPT-6 Sol and GPT-6 Luna: October 2026 updatechecked 2026-10-19; printed on 2 more pages-
GPT-6 AstraOpenAI, released 2026-09-030.998OpenAI, GPT-6 Astra System Cardchecked 2026-10-25; printed on 1 more page-
GPT-5.6 LunaOpenAI, released 2026-07-090.946OpenAI, GPT-6.1 Sol system card addendumchecked 2026-10-19; printed on 2 more pages-
GPT-5.6 SolOpenAI, released 2026-07-090.943OpenAI, GPT-6.1 Sol system card addendumchecked 2026-10-19; printed on 2 more pages-
GPT-5.5OpenAI, released 2026-04-230.958OpenAI, GPT-6.1 Sol system card addendumchecked 2026-10-19; printed on 1 more page-
GPT-5.5 InstantOpenAI0.969OpenAI, GPT-6 Sol and GPT-6 Luna: October 2026 updatechecked 2026-10-19Model as printed: GPT-5.5 Instant (May)
0.969OpenAI, GPT-6 Sol and GPT-6 Luna: October 2026 updatechecked 2026-10-19Model as printed: GPT-5.5 Instant (June)

Benign Not overrefuse

ModelValuePrinted bySetup as printed
GPT-6.1 TerraOpenAI, released 2026-10-260.982OpenAI, GPT-6.1 Terra system card addendumchecked 2026-10-26-
GPT-6.1 SolOpenAI, released 2026-09-290.982OpenAI, GPT-6.1 Sol system card addendumchecked 2026-10-19-
GPT-6 LunaOpenAI, released 2026-09-220.958OpenAI, GPT-6.1 Sol system card addendumchecked 2026-10-19; printed on 1 more page-
GPT-6 SolOpenAI, released 2026-09-220.964OpenAI, GPT-6.1 Sol system card addendumchecked 2026-10-19; printed on 1 more page-
GPT-6 AstraOpenAI, released 2026-09-030.978OpenAI, GPT-6 Astra System Cardchecked 2026-10-25; printed on 1 more page-
GPT-5.6 LunaOpenAI, released 2026-07-090.989OpenAI, GPT-6.1 Sol system card addendumchecked 2026-10-19; printed on 1 more page-
GPT-5.6 SolOpenAI, released 2026-07-090.989OpenAI, GPT-6.1 Sol system card addendumchecked 2026-10-19; printed on 1 more page-
GPT-5.5OpenAI, released 2026-04-230.917OpenAI, GPT-6.1 Sol system card addendumchecked 2026-10-19; printed on 1 more page-

Every number here is read from the same catalog file as the benchmark table, data version 2026103001, and is the number the table shows for that model and test. Where one page printed a model under several setups, each value is listed. The catalog is public in armanobosyan/sugra-llm-benchmarks; a wrong number can be reported from the table, with Report an error under the value, or as an issue in that repository.