GPQA Diamond — the 198-question expert-validated subset of GPQA (Graduate-Level Google-Proof QA); the field's standard reporting. Pool decontaminated 2026-07-03 (AA + diamond-raw + proven re-reports only).
| # | Model | Vendor | Best score | Runs | Last seen |
|---|---|---|---|---|---|
| 1 | Grok 4.6 | SpaceXAI | 94.9 | 4 | 2026-08-22 |
| 2 | Claude Mythos 5 | Anthropic | 94.6 | 1 | 2026-08-23 |
| 3 | GPT-5.6 Sol | OpenAI | 94.6 | 10 | 2026-08-23 |
| 4 | Gemini 3.7 Flash | 94.5 | 3 | 2026-08-14 | |
| 5 | GPT-5.4 Pro | OpenAI | 94.4 | 2 | 2026-06-21 |
| 6 | Gemini 3.1 Pro | 94.3 | 19 | 2026-08-24 | |
| 7 | Claude Opus 4.7 | Anthropic | 94.2 | 6 | 2026-08-23 |
| 8 | Opus 5 | Anthropic | 93.7 | 5 | 2026-07-25 |
| 9 | GPT-5.5 | OpenAI | 93.6 | 9 | 2026-08-23 |
| 10 | Claude Opus 4.8 | Anthropic | 93.6 | 7 | 2026-08-23 |
| 11 | Qwen3.8 2.4T A95B | Alibaba | 93.5 | 1 | 2026-08-14 |
| 12 | Kimi K3 | Moonshot | 93.5 | 5 | 2026-08-23 |
| 13 | GPT-5.2 Pro | OpenAI | 93.2 | 1 | 2026-08-23 |
| 14 | Grok4.5 | SpaceXAI | 93.1 | 2 | 2026-08-23 |
| 15 | MiniMax M3 | MiniMax | 93 | 4 | 2026-07-17 |
| 16 | GPT-5.4 | OpenAI | 93 | 13 | 2026-08-23 |
| 17 | GPT-5.6 Terra | OpenAI | 92.9 | 7 | 2026-08-23 |
| 18 | Gemini 3.6 Flash | 92.8 | 1 | 2026-07-22 | |
| 19 | Qwen3.8 Max Preview | Alibaba | 92.7 | 3 | 2026-08-23 |
| 20 | Claude Fable 5 | Anthropic | 92.6 | 4 | 2026-08-12 |
| 21 | GPT-5.2 | OpenAI | 92.4 | 8 | 2026-08-24 |
| 22 | Qwen3.7 Max | Alibaba | 92.4 | 5 | 2026-08-23 |
| 23 | GPT-5.6 Luna | OpenAI | 92.3 | 9 | 2026-08-23 |
| 24 | Gemini 3.5 Flash | 92.2 | 4 | 2026-07-07 | |
| 25 | Gemini 3 Pro | 91.9 | 12 | 2026-08-24 | |
| 26 | GLM-5.3 | Z.ai | 91.7 | 1 | 2026-08-19 |
| 27 | GPT-5.3 Codex | OpenAI | 91.5 | 1 | 2026-05-02 |
| 28 | Claude Opus 4.6 | Anthropic | 91.3 | 15 | 2026-08-24 |
| 29 | GLM-5.2 Full Open Source | Z.ai | 91.2 | 8 | 2026-08-23 |
| 30 | Grok 4.20 0309 V2 | SpaceXAI | 91.1 | 3 | 2026-06-18 |
| 31 | Claude Sonnet 5 | Anthropic | 91.1 | 2 | 2026-07-02 |
| 32 | Kimi K2.6 | Moonshot | 91.1 | 10 | 2026-08-23 |
| 33 | Kimi K2.6 1T A32B | Moonshot | 91 | 1 | 2026-08-24 |
| 34 | DeepSeek-V4-Flash | DeepSeek | 90.8 | 11 | 2026-08-01 |
| 35 | DeepSeek-V4-Pro | DeepSeek | 90.5 | 16 | 2026-08-23 |
| 36 | Qwen3.8 27B | Alibaba | 90.5 | 6 | 2026-08-23 |
| 37 | Muse Spark 1.2 | Meta | 90.4 | 1 | 2026-08-06 |
| 38 | Gemini 3 Flash Preview | 90.4 | 3 | 2026-08-23 | |
| 39 | Qwen3.6 Plus | Alibaba | 90.4 | 5 | 2026-08-23 |
| 40 | Hy3-preview | Tencent | 90.4 | 3 | 2026-08-23 |
| 41 | Qwen3.7 Plus Preview | Alibaba | 90.3 | 4 | 2026-08-23 |
| 42 | Grok 4.3 | SpaceXAI | 90.1 | 5 | 2026-05-23 |
| 43 | Claude Sonnet 4.6 | Anthropic | 89.9 | 4 | 2026-08-23 |
| 44 | GPT-5.2 Codex | OpenAI | 89.9 | 1 | 2026-05-02 |
| 45 | Muse Spark 1.1 | Meta | 89.8 | 1 | 2026-07-11 |
| 46 | Kimi K2.7 Code | Moonshot | 89.6 | 2 | 2026-07-07 |
| 47 | Inkling-Small | Thinking Machines | 89.5 | 4 | 2026-08-23 |
| 48 | Muse Spark | Meta | 89.5 | 3 | 2026-08-23 |
| 49 | Grok Build 0.1 | SpaceXAI | 89.5 | 2 | 2026-07-07 |
| 50 | Qwen3.5 397B A17B | Alibaba | 89.3 | 6 | 2026-08-23 |
| 51 | Nex-N2-Pro | Nex AGI | 89.2 | 2 | 2026-07-07 |
| 52 | Solar Pro4 | Upstage | 89.1 | 2 | 2026-08-23 |
| 53 | Qwen3.6 Max Preview | Alibaba | 88.8 | 1 | 2026-05-02 |
| 54 | Grok 4.20 | SpaceXAI | 88.5 | 1 | 2026-05-02 |
| 55 | Inkling | Thinking Machines | 88.3 | 4 | 2026-07-30 |
| 56 | GPT-5.1 Instant | OpenAI | 88.1 | 1 | 2026-08-23 |
| 57 | GPT-5.1 | OpenAI | 88.1 | 5 | 2026-08-23 |
| 58 | GPT-5.4 mini | OpenAI | 88 | 4 | 2026-08-23 |
| 59 | Kimi K2.5 | Moonshot | 87.9 | 9 | 2026-08-23 |
| 60 | Qwen3.6 27B | Alibaba | 87.8 | 5 | 2026-08-23 |
| 61 | Grok 4 | SpaceXAI | 87.7 | 4 | 2026-08-23 |
| 62 | Agnes 2.5 Pro Alpha | Sapiens AI | 87.6 | 1 | 2026-07-25 |
| 63 | Grok 4 Fast | SpaceXAI | 87.5 | 4 | 2026-08-23 |
| 64 | MiniMax M2.7 | MiniMax | 87.4 | 6 | 2026-07-30 |
| 65 | GPT-5 | OpenAI | 87.3 | 11 | 2026-08-23 |
| 66 | Qwen 3.5 397B 17B | Alibaba | 87.1 | 1 | 2026-08-24 |
| 67 | DeepSeek-V3.2-Speciale | DeepSeek | 87.1 | 2 | 2026-07-07 |
| 68 | Nemotron 3 Ultra 550B A55B | NVIDIA | 87 | 2 | 2026-06-06 |
| 69 | nvidia-nemotron-3-ultra-550b-a55b | NVIDIA | 87 | 1 | 2026-08-24 |
| 70 | Claude Opus 4.5 | Anthropic | 87 | 10 | 2026-08-24 |
| 71 | MiMo V2 Pro | Xiaomi | 87 | 2 | 2026-07-07 |
| 72 | Gemini 3.1 Flash Lite Preview | 86.9 | 3 | 2026-08-23 | |
| 73 | Motif-3-Beta | Motif Technologies | 86.9 | 1 | 2026-07-21 |
| 74 | GLM-5.1 | Z.ai | 86.8 | 9 | 2026-08-23 |
| 75 | MiniMax 2.7 230B A10B | MiniMax | 86.6 | 1 | 2026-08-24 |
| 76 | Qwen3.5 122B A10B | Alibaba | 86.6 | 3 | 2026-08-23 |
| 77 | Gemini 2.5 Pro Preview 06.05 (32k think) | 86.4 | 3 | 2026-08-24 | |
| 78 | GLM-5.1-744B-A40B | Z.ai | 86.1 | 1 | 2026-08-24 |
| 79 | GLM-5 | Z.ai | 86 | 5 | 2026-05-18 |
| 80 | Qwen3.6 35B A3B | Alibaba | 86 | 6 | 2026-08-23 |
| 81 | GPT-5.1 Codex | OpenAI | 86 | 1 | 2026-05-02 |
| 82 | GLM-4.7 | Z.ai | 85.9 | 5 | 2026-08-23 |
| 83 | Qwen3.5 27B | Alibaba | 85.8 | 5 | 2026-08-23 |
| 84 | Gemma 4 31B | 85.7 | 31 | 2026-08-24 | |
| 85 | Solar Open2 250B | Upstage | 85.7 | 1 | 2026-08-13 |
| 86 | A.X-K2 | SK Telecom | 85.7 | 1 | 2026-08-13 |
| 87 | GPT-5.5 Instant | OpenAI | 85.6 | 5 | 2026-08-23 |
| 88 | Ling-3.0-flash | InclusionAI | 85.5 | 1 | 2026-08-06 |
| 89 | MiMo V2 Omni | Xiaomi | 85.5 | 4 | 2026-07-07 |
| 90 | Grok 4.1 Fast | SpaceXAI | 85.3 | 2 | 2026-06-23 |
| 91 | MiniMax M2.5 | MiniMax | 85.2 | 3 | 2026-08-24 |
| 92 | ERNIE 5.0 Thinking Preview | Baidu | 85 | 2 | 2026-08-23 |
| 93 | MiMo V2.5 Pro | Xiaomi | 84.9 | 7 | 2026-08-23 |
| 94 | Nanbeige4.1-3B | Nanbeige | 84.9 | 2 | 2026-07-07 |
| 95 | Claude 3.7 Sonnet | Anthropic | 84.8 | 3 | 2026-08-23 |
| 96 | GLM-5-Turbo | Z.ai | 84.7 | 2 | 2026-07-07 |
| 97 | MiMo V2 Flash | Xiaomi | 84.6 | 7 | 2026-08-23 |
| 98 | Grok 3 | SpaceXAI | 84.6 | 4 | 2026-08-23 |
| 99 | MAI-Code-1-Flash | Microsoft | 84.6 | 1 | 2026-08-23 |
| 100 | o3-pro | OpenAI | 84.5 | 2 | 2026-07-07 |
Best tracked score per model (default configuration; source-attributed and verification-tiered). Open a model for its full benchmark surface, provenance and pricing.