Best for reasoning

Ranked by GPQA Diamond - graduate-level science questions written to be hard to look up. Source: Epoch AI. Variants (reasoning effort, batch) are folded into one row.

#	Model	Capability	AA index	Reasoning	Math	Coding	Context	$/M	Weights	Released
201	MMinistral 3BMistral AI	118.1	—	25.3	—	—	—	—	Closed	16 Oct 2024
202	DDeepSeek LLM 67BDeepSeek	110.5	—	24.6	0.8	—	—	—	Open	29 Nov 2023
203	?granite-4.0-1b	—	—	24.0	0.8	—	—	—	—	28 Oct 2025
204	MLlama 3.2 1BMeta	102.0	—	23.9	0.6	—	—	—	Open	24 Sep 2024
205	GGemma 3 4BGoogle	116.0	—	23.2	7.5	—	131K	$0.063	Open	12 Mar 2025
206	GGemma 3 1BGoogle	—	—	19.9	1.1	—	—	—	Open	12 Mar 2025
207	MMistral 7B v0.3Mistral AI	108.7	—	15.2	0.3	—	—	—	Open	27 May 2024
208	0Yi-34B01.AI	117.3	—	14.7	—	—	—	—	Open	2 Nov 2023
209	?granite-4.0-350m	—	—	11.2	0.3	—	—	—	—	28 Oct 2025
210	?deepseek-r1-0528-qwen3-8b	—	—	9.3	43.9	—	—	—	—	28 May 2025

• OGPT-6.1 Sol
  OpenAI · 29 Sep 2026 · AA 51.8 · 1.05M · $4.00/M
• OGPT-6.1 Sol Pro
  OpenAI · 29 Sep 2026 · 1.05M · $4.00/M
• AClaude Sonnet 5.5
  Anthropic · 28 Sep 2026 · AA 56 · 1M · $4.00/M
• AQwen3.8 Max Prime
  Alibaba (Qwen) · 23 Sep 2026 · 1M · $6.00/M
• OGPT-6 Sol
  OpenAI · 22 Sep 2026 · AA 47.5 · 1.05M · $4.00/M
• OGPT-6 Luna
  OpenAI · 22 Sep 2026 · AA 37.3 · 1.05M · $0.20/M
• OGPT-6 Sol Pro
  OpenAI · 22 Sep 2026 · 1.05M · $4.00/M
• OGPT-6 Luna Pro
  OpenAI · 22 Sep 2026 · 1.05M · $0.20/M

Not ranked until an independent evaluator (Epoch AI) publishes scores. We don't use vendor-reported benchmark claims for rankings.