The highest-quality subset of GPQA, designed to evaluate expert-level knowledge and reasoning across difficult scientific domains.
Browse the latest scores, model modes, release dates, and parameter sizes for GPQA Diamond.
Data sourced primarily from official releases (GitHub, Hugging Face, papers), then benchmark leaderboards, then third-party evaluators. Learn about our data methodology
2 parallel-mode results hidden
| Rank | Model | License | |||
|---|---|---|---|---|---|
 Claude Mythos Preview Extended Thinking | 94.60 | 2026-04-07 | Unknown | Closed | |
 GPT-5.4 Pro Thinking Level · High | 94.40 | 2026-03-05 | Unknown | Closed | |
 Gemini 3.1 Pro Preview Thinking Level · High | 94.30 | 2026-02-20 | Unknown | Closed | |
4 | Opus 4.7 Extended Thinking | 94.20 | 2026-04-16 | Unknown | Closed |
5 | GPT-5.5 Thinking Level · High | 93.60 | 2026-04-23 | Unknown | Closed |
6 | Claude Opus 4.8 Thinking Level · High | 93.60 | 2026-05-28 | Unknown | Closed |
7 |  Kimi K3 Thinking Level · Max | 93.50 | 2026-07-16 | 2800B | Free Commercial |
8 | GPT-5.2 Deep Thinking Mode | 93.20 | 2025-12-11 | Unknown | Closed |
9 | GPT-5.2 Pro Thinking Enabled | 93.20 | 2025-12-11 | Unknown | Closed |
10 | GPT-5.4 Thinking Level · Extra High | 92.80 | 2026-03-05 | Unknown | Closed |
11 | GPT-5.2 Thinking Level · Extra High | 92.40 | 2025-12-11 | Unknown | Closed |
12 |  Qwen3.7-Max-Preview Thinking Level · Max | 92.40 | 2026-05-20 | Unknown | Closed |
13 | Gemini 3.0 Pro (Preview 11-2025) Thinking Enabled | 91.90 | 2025-11-18 | Unknown | Closed |
14 | Claude Opus 4.6 Extended Thinking | 91.31 | 2026-02-05 | Unknown | Closed |
15 |  GLM-5.2 Thinking Enabled | 91.20 | 2026-06-13 | 753.3B | Free Commercial |
16 | Gemini 3.0 Pro (Preview 11-2025) Thinking Level · High | 91.00 | 2025-11-18 | Unknown | Closed |
17 | Kimi K2.6 Thinking Enabled | 90.50 | 2026-04-20 | 1000B | Free Commercial |
18 | Gemini 3.0 Flash Thinking Enabled | 90.40 | 2025-12-17 | Unknown | Closed |
19 | Qwen 3.6 Plus Preview Thinking Enabled | 90.40 | 2026-03-31 | Unknown | Closed |
20 | Qwen3.6-Max-Preview Thinking Level · Max | 90.40 | 2026-04-18 | Unknown | Closed |
21 |  Hy3 Thinking Level · High | 90.40 | 2026-07-06 | 295B | Free Commercial |
22 |  DeepSeek-V4-Pro Thinking Level · Max | 90.10 | 2026-04-24 | 1600B | Free Commercial |
23 | Claude Sonnet 4.6 Thinking Enabled | 89.90 | 2026-02-17 | Unknown | Closed |
24 |  Muse Spark Thinking Enabled | 89.50 | 2026-04-08 | Unknown | Closed |
25 | GPT-5-Pro Thinking EnabledTools | 89.40 | 2025-08-07 | Unknown | Closed |
26 | DeepSeek-V4-Pro Thinking Level · High | 89.10 | 2026-04-24 | 1600B | Free Commercial |
27 | GPT-5-Pro Thinking Enabled | 88.40 | 2025-08-07 | Unknown | Closed |
28 | Qwen3.5-397B-A17B Thinking Enabled | 88.40 | 2026-02-16 | 39.7B | Free Commercial |
