GPQA Leaderboard

Graduate-level science questions, "Google-proof". · Prices checked 12 August 2026

GPQA Diamond is a set of graduate-level questions in biology, physics and chemistry, written so that searching the web does not help. It isolates reasoning from recall, which makes it the cleanest signal we track for hard analytical work.

Reasoning models tend to lead this table, and they also tend to be the most expensive per token, because the internal chain of thought they generate before answering is billed as output.

Leaderboard

60 of 130 scored

Claude Fable 5 leads on GPQA Diamond at 95%. The median across the 130 models we hold a score for is 51%. The cheapest model on this board is Phi-4 at $0.133/M blended, scoring 56%.

#	Model	Provider	Score	Input /1M	Output /1M	Blended 70/30	Context
1	Claude Fable 5	Anthropic	95%	$10.00	$50.00	$22.00	1,000,000
2	GPT-5.6 Sol	OpenAI	94.6%	$5.00	$30.00	$12.50	1,050,000
3	Claude Opus 4.8	Anthropic	94%	$5.00	$25.00	$11.00	1,000,000
4	GPT-5.6 Terra	OpenAI	92.9%	$2.00	$12.00	$5.00	1,050,000
5	GPT-5.6 Luna	OpenAI	92.3%	$0.200	$1.20	$0.500	1,050,000
6	o3	OpenAI	88%	$10.00	$40.00	$19.00	200,000
7	GPT-5	OpenAI	84%	$1.25	$10.00	$3.88	400,000
8	Gemini 2.5 Pro	Google	84%	$1.25	$10.00	$3.88	1,000,000
9	o4-mini	OpenAI	81%	$1.10	$4.40	$2.09	200,000
10	Claude Opus 4.6	Anthropic	79%	$5.00	$25.00	$11.00	1,000,000
11	o3-mini	OpenAI	79%	$1.10	$4.40	$2.09	200,000
12	o1	OpenAI	78%	$15.00	$60.00	$28.50	200,000
13	Claude Opus 4.5	Anthropic	76%	$15.00	$75.00	$33.00	200,000
14	Claude Sonnet 4.6	Anthropic	75%	$3.00	$15.00	$6.60	1,000,000
15	Grok 3	xAI	75%	$3.00	$15.00	$6.60	131,000
16	Grok 3 Fast	xAI	75%	$5.00	$25.00	$11.00	131,000
17	Claude Sonnet 4.5	Anthropic	73%	$3.00	$15.00	$6.60	200,000
18	DeepSeek-R1-0528	DeepSeek	73%	$0.550	$2.19	$1.04	64,000
19	GPT-4.1	OpenAI	71%	$2.00	$8.00	$3.80	1,000,000
20	DeepSeek-R1	DeepSeek	71%	$0.550	$2.19	$1.04	64,000
21	DeepSeek-R1 (Groq)	Groq	71%	$0.750	$0.990	$0.822	128,000
22	DeepSeek-R1 (Together)	Together AI	71%	$3.00	$7.00	$4.20	64,000
23	DeepSeek-R1 (fw)	Fireworks AI	71%	$3.00	$8.00	$4.50	64,000
24	DeepSeek-R1 (DI)	DeepInfra	71%	$0.550	$2.19	$1.04	64,000
25	DeepSeek-R1 (Cerebras)	Cerebras	71%	$0.550	$0.990	$0.682	64,000
26	Claude Sonnet 3.7	Anthropic	70%	$3.00	$15.00	$6.60	200,000
27	Gemini 2.5 Flash	Google	70%	$0.150	$0.600	$0.285	1,000,000
28	Sonar Reasoning Pro	Perplexity	70%	$2.00	$8.00	$3.80	127,000
29	Kimi k1.5	Moonshot AI	70%	$2.50	$10.00	$4.75	128,000
30	Llama 4 Maverick	Meta	67%	$0.200	$0.600	$0.320	128,000
31	Claude Sonnet 3.5	Anthropic	65%	$3.00	$15.00	$6.60	200,000
32	DeepSeek-R1-Distill-70B	DeepSeek	65%	$0.350	$0.880	$0.509	128,000
33	Sonar Deep Research	Perplexity	65%	$2.00	$8.00	$3.80	127,000
34	QwQ-32B	Qwen	65%	$0.600	$2.40	$1.14	131,000
35	Grok 3 mini	xAI	64%	$0.300	$0.500	$0.360	131,000
36	Gemini 2.0 Flash	Google	62%	$0.100	$0.400	$0.190	1,000,000
37	DeepSeek-R1-Distill-32B	DeepSeek	62%	$0.200	$0.500	$0.290	128,000
38	GPT-4.1 mini	OpenAI	60%	$0.400	$1.60	$0.760	1,000,000
39	DeepSeek-V3-0324	DeepSeek	60%	$0.270	$1.10	$0.519	128,000
40	Nova Premier	Amazon	60%	$2.50	$12.50	$5.50	300,000
41	Sonar Pro	Perplexity	60%	$3.00	$15.00	$6.60	200,000
42	Sonar Reasoning	Perplexity	60%	$1.00	$5.00	$2.20	127,000
43	Llama-3.1-Nemotron-Ultra-253B	Nvidia	60%	$1.60	$1.60	$1.60	128,000
44	Qwen2.5-Max	Qwen	60%	$1.60	$6.40	$3.04	32,000
45	Gemini 1.5 Pro	Google	59%	$1.25	$5.00	$2.38	2,000,000
46	DeepSeek-V3	DeepSeek	59%	$0.270	$1.10	$0.519	64,000
47	Llama 4 Scout	Meta	57%	$0.100	$0.350	$0.175	512,000
48	Claude Haiku 4.5	Anthropic	56%	$0.800	$4.00	$1.76	200,000
49	Grok 2	xAI	56%	$2.00	$10.00	$4.40	131,000
50	Grok 2 Vision	xAI	56%	$2.00	$10.00	$4.40	32,000
51	Phi-4	Microsoft	56%	$0.070	$0.280	$0.133	16,000
52	GPT-4o	OpenAI	53%	$2.50	$10.00	$4.75	128,000
53	Command A	Cohere	53%	$2.50	$10.00	$4.75	256,000
54	Mistral Large 3	Mistral	52%	$0.500	$1.50	$0.800	256,000
55	Mistral Large 2	Mistral	52%	$2.00	$6.00	$3.20	128,000
56	Mistral-Large-2 (NIM)	Nvidia	52%	$2.00	$6.00	$3.20	128,000
57	Gemini 2.0 Flash Lite	Google	51%	$0.075	$0.300	$0.142	1,000,000
58	Gemini 1.5 Flash	Google	51%	$0.075	$0.300	$0.142	1,000,000
59	DeepSeek-R1-Distill-14B	DeepSeek	51%	$0.100	$0.350	$0.175	128,000
60	Llama 3.1 405B	Meta	51%	$0.800	$0.800	$0.800	128,000

About these scores

Graduate-level science questions, "Google-proof".
Scores are as published by the model’s provider or a public leaderboard. Provider-reported figures are self-reported and are not independently re-run by us.
Where a hosted variant of an open-weights model has no score of its own, it inherits the base model’s — the weights are the same.
Prices beside each score are the provider’s published token rates. How the blended rate is calculated.

Showing the top 60 of 130 models we hold a GPQA Diamond score for.

Frequently asked questions

What does GPQA measure?

Multi-step scientific reasoning at graduate level. The questions are deliberately "Google-proof": domain experts answer roughly two-thirds correctly even with unrestricted web access, so a high score reflects genuine reasoning rather than retrieval or memorisation.

Why do reasoning models score higher on GPQA?

They are trained to work through a problem step by step before answering, which is exactly what these questions require. That extra deliberation costs output tokens, so the models at the top of this table are usually also near the top of the price sheet.

Other leaderboards

SWE-bench Leaderboard MMLU Leaderboard MATH Leaderboard HumanEval Leaderboard LMArena Leaderboard

Rankings that weigh price too

Best value AI models Best AI models for coding Best reasoning models Compare models side by side

Leaderboards are recomputed on every deploy from the catalogue. Prices checked 12 August 2026.