ModelsCompareBest forBenchmarksStatusPricingAPI

What is GPQA?

Graduate-level, 'Google-proof' multiple-choice questions in biology, physics, and chemistry written by domain experts to be hard to answer even with web search.

GPQA scores by model

1
GPT-6 AstraOpenAI96% ↗
2
Gemini 3.8 FlashGoogle95.4% ↗
3
Gemini 3.7 FlashGoogle94.8% ↗
4
GPT-5.6 SolOpenAI94.6% ↗
5
GPT-5.4 ProOpenAI94.6% ↗
6
Gemini 3.1 Pro PreviewGoogle94.4% ↗
7
Claude Opus 4.7Anthropic94.2% ↗
8
Gemini 3.6 FlashGoogle94.1% ↗
9
Grok 4.6xAI94% ↗
10
Claude Opus 5Anthropic93.9% ↗
11
Claude Opus 4.8Anthropic93.6% ↗
12
Kimi K3Moonshot AI93.5% ↗
13
Grok 4.5xAI93.4% ↗
14
GPT-5.6 TerraOpenAI93.3% ↗
15
GPT-5.4OpenAI93.3% ↗
16
GPT-5.2 ProOpenAI93.2% ↗
17
Gemini 3.5 FlashGoogle92.8% ↗
18
Qwen3.8-MaxQwen92.7% ↗
19
Qwen3.8 2.4T-A95BQwen92.6% ↗
20
GPT-5.2OpenAI92.4% ↗
21
Dola Seed 2.0 ProByteDance Seed92.4% ↗
22
GPT-5.6 LunaOpenAI92.3% ↗
23
GLM-5.2Z.ai91.9% ↗
24
Qwen3.8-Flash-NextQwen91.7% ↗
25
Claude Opus 4.6Anthropic91.3% ↗
26
DeepSeek V4.1 FlashDeepSeek90.9% ↗
27
DeepSeek V4 ProDeepSeek90.9% ↗
28
MiniMax M3MiniMax90.9% ↗
29
GLM-5.3Z.ai90.9% ↗
30
Kimi K2.6Moonshot AI90.8% ↗
31
GPT-5.5OpenAI90.7% ↗
32
Claude Sonnet 5Anthropic90.5% ↗
33
Hy3Tencent Hunyuan90.4% ↗
34
Gemini 3 Flash PreviewGoogle90.4% ↗
35
GLM-5.3-FlashZ.ai90.2% ↗
36
Claude Sonnet 4.6Anthropic89.9% ↗
37
GLM-5.1Z.ai89.9% ↗
38
Grok 4.20 ReasoningxAI89.3% ↗
39
Qwen3.8 27BQwen89.2% ↗
40
LongCat 2.0Meituan LongCat88.9% ↗