Models · Edition 1.0

Models tested in edition 1.0.

18 models have an official result in this edition. Each summary shows the model's results by subject and links to every round.

For the ranked comparison, see the results overview.

Model	Rank	Question score	Rounds solved
Claude Fable 5High	Rank 1	12.06	35 of 35
Claude Fable 5.1High	Rank 2	12.11	34 of 35
Claude Opus 5High	Rank 3	12.34	35 of 35
Claude Sonnet 5High	Rank 10	14.74	35 of 35
Gemini 3.6 FlashHigh	Rank 14	16.89	34 of 35
Gemini 3.7 FlashHigh	Rank 7	14.03	34 of 35
Gemini 3.8 FlashHigh	Rank 5	13.14	35 of 35
GPT-5 NanoMedium	Rank 13	15.49	33 of 35
GPT-5.6 LunaHigh	Rank 16	17.74	32 of 35
GPT-5.6 SolHigh	Rank 9	14.40	35 of 35
GPT-6 AstraHigh	Rank 11	15.09	33 of 35
gpt-oss-120BHigh	Rank 6	13.63	33 of 35
Grok 4.5High	Rank 12	15.17	33 of 35
Grok 4.6High	Rank 8	14.29	35 of 35
Kimi K3High	Rank 4	12.74	33 of 35
Llama 4 MaverickNon Thinking	Rank 18	32.23	17 of 35
Mistral Medium 3.5High	Rank 17	20.06	31 of 35
Ox AlphaHigh	Rank 15	17.60	32 of 35

Lower question scores are better.