Reasoning

GPQA-Diamond

Graduate-level Physics, Chemistry, Biology - preguntas de nivel doctoral.

54 modelos publicaron score

#	Modelo	Empresa	Score
1	Claude Mythos Preview	Anthropic	94.6
2	GPT-5.4 Pro	OpenAI	94.4
3	Gemini 3.1 Pro	Google DeepMind	94.3
4	Claude Opus 4.7	Anthropic	94.2
5	GPT-5.5	OpenAI	93.6
6	GPT-5.2 Pro	OpenAI	93.2
7	GPT-5.4	OpenAI	92.8
8	GPT-5.2	OpenAI	92.4
9	Gemini 3 Pro	Google DeepMind	91.9
10	Claude Opus 4.6	Anthropic	91.3
11	Kimi K2.6	Moonshot AI	90.5
12	Gemini 3 Flash	Google DeepMind	90.4
13	DeepSeek V4 Pro	DeepSeek	90.1
14	Doubao Seed 2.0 Pro	ByteDance	88.9
15	GPT-5.4 mini	OpenAI	88.0
16	Grok 4 Heavy	xAI	88.0
17	Grok 4	xAI	88.0
18	Kimi K2.5	Moonshot AI	87.6
19	Qwen3-Max-Thinking	Alibaba	87.4
20	Claude Opus 4.5	Anthropic	87.0
21	MiMo V2 Pro	Xiaomi	87.0
22	Gemini 3.1 Flash-Lite	Google DeepMind	86.9
23	Qwen3.6-35B-A3B	Alibaba	86.0
24	GLM-5	Zhipu AI	86.0
25	DeepSeek V3.2 Speciale	DeepSeek	85.7
26	MiniMax M2.5	MiniMax	85.2
27	Gemma 4 (31B dense)	Google DeepMind	84.3
28	MiMo V2 Flash	Xiaomi	83.5
29	GLM-4.6	Zhipu AI	82.9
30	GPT-5.4 nano	OpenAI	82.8
31	DeepSeek V3.2	DeepSeek	82.4
32	Gemma 4 26B-A4B	Google DeepMind	82.3
33	DeepSeek R1 0528	DeepSeek	81.0
34	EXAONE 4.5 33B	LG AI Research	80.5
35	Nova 2 Lite	Amazon	79.6
36	Nemotron 3 Super	Nvidia	79.4
37	K-EXAONE 236B-A23B	LG AI Research	79.1
38	Qwen3-Max	Alibaba	76.4
39	Magistral Medium 1.2	Mistral AI	76.3
40	Claude Sonnet 4.6	Anthropic	74.1
41	Llama 4 Behemoth	Meta	73.7
42	Step-3	StepFun	73.0
43	Nemotron 3 Nano	Nvidia	73.0
44	Mistral Small 4	Mistral AI	71.2
45	Llama 4 Maverick	Meta	69.8
46	Gemma 4 E4B	Google DeepMind	58.6
47	Llama 4 Scout	Meta	57.2
48	Reka Flash 3	Reka	52.9
49	Yi-Lightning	01.AI	50.9
50	Command A	Cohere	50.8
51	Nova Pro	Amazon	46.9
52	Mistral Large 3	Mistral AI	43.9
53	Gemma 4 E2B	Google DeepMind	43.4
54	Jamba 1.7 Large	AI21 Labs	39.0

← Todos los benchmarks Como medimos