WeirdML v3

Unusual machine-learning tasks the model solves by writing and running code.

Results

#	Model	Provider	Result	Date
1	GPT-6 Astra Extra High	OpenAI	42.2%	2026-11-11
2	GPT-6.1 Sol Extra High	OpenAI	38.3%	2026-11-11
3	Claude Opus 5.5 Extra High	Anthropic	31.2%	2026-11-11
4	GPT-6.1 Terra Extra High	OpenAI	28.1%	2026-11-11
5	Claude Fable 5.1 Extra High	Anthropic	26.0%	2026-11-11
6	Gemini 4 Argon High	Google	20.3%	2026-11-11
7	Claude Sonnet 5.5 Extra High	Anthropic	19.9%	2026-11-11
8	GPT-6 Sol Extra High	OpenAI	19.7%	2026-11-11
9	Claude Opus 5 Extra High	Anthropic	19.0%	2026-11-11
10	Claude Fable 5 Extra High	Anthropic	15.6%	2026-11-11
11	Muse Wonder Extra High	Meta	15.2%	2026-11-11
12	GPT-5.6 Sol Extra High	OpenAI	15.0%	2026-11-11
13	Claude Haiku 5.5 Extra High	Anthropic	11.7%	2026-11-11
14	Kimi K3.1 Pro Extra High	Moonshot AI	9.9%	2026-11-11
15	DeepSeek V4.1 Flash 1017 High	DeepSeek	9.2%	2026-11-11
16	Grok 4.7 Extra High	xAI	8.5%	2026-11-11
17	GPT-6 Luna Extra High	OpenAI	7.9%	2026-11-11
18	Kimi K3 High	Moonshot AI	7.5%	2026-11-11
19	Gemini 3.8 Flash High	Google	7.4%	2026-11-11
20	Muse Spark 1.3 Extra High	Meta	6.7%	2026-11-11
21	Claude Opus 4.5 High	Anthropic	6.5%	2026-11-11
22	DeepSeek V4.1 Flash High	DeepSeek	6.0%	2026-11-11
23	GPT-5.6 Luna Extra High	OpenAI	5.2%	2026-11-11
24	GPT-5.2 Extra High	OpenAI	5.0%	2026-11-11
25	Mistral Large 4 Preview High	Mistral	4.5%	2026-11-11
26	Inkling Extra High	Thinking Machines	3.9%	2026-11-11