🧠 Reasoning LLMリーダーボード

Mathematical reasoning, science, and logical problem solving.

All Chat Code Reasoning Image Video Agent

#	モデル	総合	MMLU	GPQA	AIME	MATH	速度
1	Claude Opus 4 Anthropic	97.8	92.1%	83.5%	87.4%	96.1%	45 t/s
2	Gemini 2.5 Pro Google	97.5	92%	84%	92%	97.2%	85 t/s
3	o3 OpenAI	97.1	91.2%	81.3%	91.6%	96.7%	58 t/s
4	DeepSeek R1 DeepSeek	92.3	90.8%	71.5%	79.8%	97.3%	32 t/s
5	o4-mini OpenAI	91.5	88.3%	72.1%	79.4%	93.2%	145 t/s
6	Claude Sonnet 4 Anthropic	90.3	90.4%	70.2%	40.5%	85.6%	79 t/s
7	GPT-4o OpenAI	89.2	88.7%	53.6%	9.3%	76.6%	110 t/s
8	Grok 3 xAI	88.8	89.2%	68.5%	55.3%	88.1%	55 t/s
9	Claude 3.5 Sonnet Anthropic	88.4	88.7%	65%	16%	78.3%	82 t/s
10	Llama 4 Maverick Meta	87.5	89.5%	62.8%	45.3%	84.2%	120 t/s
11	DeepSeek V3 DeepSeek	86.1	88.5%	59.1%	39.2%	90.2%	68 t/s
12	Qwen 3 235B Alibaba	84.5	87.8%	58.3%	35.6%	88.5%	95 t/s
13	Llama 3.1 405B Meta	82.1	87.3%	51.1%	7.8%	73.8%	29 t/s
14	Llama 3.3 70B Meta	79.8	86%	48.7%	8.2%	72.1%	65 t/s
15	Mistral Large Mistral AI	76.5	84%	46.5%	5.6%	69.2%	48 t/s
16	Qwen 2.5 72B Alibaba	76	85.2%	45.6%	6.8%	71.5%	55 t/s
17	Phi-4 Microsoft	64.5	78.5%	38.5%	3.8%	62.5%	125 t/s