TestsBenchmarks · Knowledge
GPQA Diamond
Accuracy on 198 graduate-level, Google-proof science multiple-choice questions.
% solvedHigher is betterToo easy now: the top models all score near perfectSaturatedThe test's websiteOfficial page ↗
Measured by independent testersIndependent Reported by the makerVendor-reported
| Thinking levelSetting | |||||||
|---|---|---|---|---|---|---|---|
| 1 | GPT-6 Astra | OpenAI | 96.3% | Extra highGPT-6 Astra (Xhigh) | Independent testIndependentArtificial Analysis ↗ | — | 1 Oct 2026 |
| 2 | Claude Sonnet 5.5 | Anthropic | 95.6% | Maxclaude-sonnet-5-5_max | Independent testIndependentEpoch AI ↗ | — | 29 Sep 2026 |
| 3 | Gemini 3.1 Pro (Preview) | Google (Gemini / DeepMind) | 95.5% | Highreasoning_effort=high | Independent testIndependentVals.ai ↗ | — | 1 Sep 2026 |
| 4 | GPT-6.1 Sol | OpenAI | 95.4% | Maxgpt-6.1-sol_max | Independent testIndependentEpoch AI ↗ | — | 29 Sep 2026 |
| 5 | Gemini 3.8 Flash | Google (Gemini / DeepMind) | 95.3% | HighGemini 3.8 Flash (High) | Independent testIndependentArtificial Analysis ↗ | — | 1 Oct 2026 |
| 6 | Grok 4.6 | SpaceXAI (formerly xAI) | 94.9% | HighGrok 4.6 (High) | Independent testIndependentArtificial Analysis ↗ | — | 1 Oct 2026 |
| 7 | GPT-5.4 Pro | OpenAI | 94.6% | Extra highgpt-5.4-pro-2026-03-05_xhigh | Independent testIndependentEpoch AI ↗ | — | 20 Mar 2026 |
| 8 | GPT-5.6 Sol | OpenAI | 94.6% | DefaultGPT-5.6 launch table (effort not stated; text cites max for headline results) | Maker's own figureVendor-reportedOpenAI ↗ | — | 9 Jul 2026 |
| 9 | Gemini 3.7 Flash | Google (Gemini / DeepMind) | 94.5% | HighGemini 3.7 Flash (High) | Independent testIndependentArtificial Analysis ↗ | — | 1 Oct 2026 |
| 10 | GPT-6 Sol | OpenAI | 94.3% | Maxgpt-6-sol_max | Independent testIndependentEpoch AI ↗ | — | 22 Sep 2026 |
| 11 | Muse Spark 1.3 | Meta (Meta Superintelligence Labs, Muse) | 94.1% | Extra highMuse Spark 1.3 (Xhigh) | Independent testIndependentArtificial Analysis ↗ | — | 1 Oct 2026 |
| 12 | GPT-5.5 Pro | OpenAI | 93.9% | Extra highgpt-5.5-pro-pre-release_xhigh | Independent testIndependentEpoch AI ↗ | — | 24 Apr 2026 |
| 13 | Claude Opus 5 | Anthropic | 93.7% | Extra highClaude Opus 5 (Adaptive Reasoning, Xhigh Effort) | Independent testIndependentArtificial Analysis ↗ | — | 1 Oct 2026 |
| 14 | Claude Fable 5.1 | Anthropic | 93.7% | MaxClaude Fable 5.1 (Adaptive Reasoning, Max Effort, Default Fallback) | Independent testIndependentArtificial Analysis ↗ | — | 1 Oct 2026 |
| 15 | GPT-5.5 | OpenAI | 93.6% | DefaultGPT-5.6 launch table (effort not stated; text cites max for headline results) | Maker's own figureVendor-reportedOpenAI ↗ | — | 9 Jul 2026 |
| 16 | Qwen3.8 2.4T A95B (open weights) | Qwen (Alibaba) | 93.5% | DefaultQwen3.8 2.4T A95B | Independent testIndependentArtificial Analysis ↗ | — | 1 Oct 2026 |
| 17 | Kimi K3 | Moonshot AI (Kimi) | 93.5% | MaxKimi K3 (Max) | Independent testIndependentArtificial Analysis ↗ | — | 1 Oct 2026 |
| 18 | Grok 4.5 | SpaceXAI (formerly xAI) | 93.1% | HighGrok 4.5 (High) | Independent testIndependentArtificial Analysis ↗ | — | 1 Oct 2026 |
| 19 | MiniMax M3 | MiniMax | 92.9% | DefaultMiniMax-M3 | Independent testIndependentArtificial Analysis ↗ | — | 1 Oct 2026 |
| 20 | GPT-5.6 Terra | OpenAI | 92.9% | DefaultGPT-5.6 launch table (effort not stated; text cites max for headline results) | Maker's own figureVendor-reportedOpenAI ↗ | — | 9 Jul 2026 |
| 21 | Gemini 3.6 Flash | Google (Gemini / DeepMind) | 92.8% | HighGemini 3.6 Flash (High) | Independent testIndependentArtificial Analysis ↗ | — | 1 Oct 2026 |
| 22 | Qwen3.8 Max (0902) | Qwen (Alibaba) | 92.8% | DefaultQwen3.8 Max (0902) | Independent testIndependentArtificial Analysis ↗ | — | 1 Oct 2026 |
| 23 | DeepSeek V4 Pro (0813) | DeepSeek | 92.8% | MaxDeepSeek V4 Pro 0813 (Max) | Independent testIndependentArtificial Analysis ↗ | — | 1 Oct 2026 |
| 24 | Qwen3.8 Max (0803) | Qwen (Alibaba) | 92.7% | Extra highqwen3.8-max_xhigh | Independent testIndependentEpoch AI ↗ | — | 4 Aug 2026 |
| 25 | Gemini 3.5 Flash | Google (Gemini / DeepMind) | 92.7% | Highreasoning_effort=high | Independent testIndependentVals.ai ↗ | — | 1 Sep 2026 |
| 26 | Claude Fable 5 | Anthropic | 92.6% | MaxClaude Fable 5 (Adaptive Reasoning, Max Effort, Opus 4.8 Fallback) | Independent testIndependentArtificial Analysis ↗ | — | 1 Oct 2026 |
| 27 | Gemini 3 Pro | Google (Gemini / DeepMind) | 92.6% | Defaultgemini-3-pro-preview | Independent testIndependentEpoch AI ↗ | — | 19 Nov 2025 |
| 28 | Qwen3.8 Max (0803) | Qwen (Alibaba) | 92.6% | Defaultthinking (default reasoning_effort=xhigh; eval setting not stated) | Maker's own figureVendor-reportedQwen (Alibaba) ↗ | — | 3 Aug 2026 |
| 29 | Qwen3.8-Flash-Next | Qwen (Alibaba) | 92.3% | DefaultQwen3.8-Flash-Next | Independent testIndependentArtificial Analysis ↗ | — | 1 Oct 2026 |
| 30 | GPT-5.6 Luna | OpenAI | 92.3% | DefaultGPT-5.6 launch table (effort not stated; text cites max for headline results) | Maker's own figureVendor-reportedOpenAI ↗ | — | 9 Jul 2026 |
| 31 | Claude Opus 4.8 | Anthropic | 92.0% | MaxClaude Opus 4.8 (Adaptive Reasoning, Max Effort) | Independent testIndependentArtificial Analysis ↗ | — | 1 Oct 2026 |
| 32 | GPT-5.4 | OpenAI | 92.0% | Extra highGPT-5.4 (Xhigh) | Independent testIndependentArtificial Analysis ↗ | — | 1 Oct 2026 |
| 33 | GLM-5.3 | Z.ai (Zhipu) | 91.7% | MaxGLM-5.3 (Max) | Independent testIndependentArtificial Analysis ↗ | — | 1 Oct 2026 |
| 34 | Qwen3.8 Flash | Qwen (Alibaba) | 91.7% | Defaultthinking (setting not stated) | Maker's own figureVendor-reportedQwen (Alibaba) ↗ | — | 26 Aug 2026 |
| 35 | GPT-5.2 | OpenAI | 91.7% | Extra highreasoning_effort=xhigh | Independent testIndependentVals.ai ↗ | — | 1 Sep 2026 |
| 36 | Grok 4.3 | SpaceXAI (formerly xAI) | 91.4% | Default | Independent testIndependentVals.ai ↗ | — | 1 Sep 2026 |
| 37 | Claude Opus 4.7 | Anthropic | 91.4% | MaxClaude Opus 4.7 (Adaptive Reasoning, Max Effort) | Independent testIndependentArtificial Analysis ↗ | — | 1 Oct 2026 |
| 38 | DeepSeek V4 Flash Vision Exp | DeepSeek | 91.3% | MaxDeepSeek V4 Flash Vision (Max) | Independent testIndependentArtificial Analysis ↗ | — | 1 Oct 2026 |
| 39 | GLM-5.3-Flash | Z.ai (Zhipu) | 91.2% | DefaultGLM 5.3 Flash | Independent testIndependentArtificial Analysis ↗ | — | 1 Oct 2026 |
| 40 | GLM-5.2 | Z.ai (Zhipu) | 91.2% | Defaultthinking (effort not stated) | Maker's own figureVendor-reportedZ.ai ↗ | — | 16 Jun 2026 |
| 41 | Claude Sonnet 5 | Anthropic | 91.1% | MaxClaude Sonnet 5 (Adaptive Reasoning, Max Effort) | Independent testIndependentArtificial Analysis ↗ | — | 1 Oct 2026 |
| 42 | DeepSeek V4.1 Flash | DeepSeek | 90.9% | Maxreasoning_effort=100 (max) | Maker's own figureVendor-reportedDeepSeek ↗ | — | 10 Sep 2026 |
| 43 | DeepSeek V4 Flash (0731) | DeepSeek | 90.8% | MaxDeepSeek V4 Flash 0731 (Max) | Independent testIndependentArtificial Analysis ↗ | — | 1 Oct 2026 |
| 44 | Qwen3.8 27B | Qwen (Alibaba) | 90.5% | Extra highQwen3.8 27B (Xhigh) | Independent testIndependentArtificial Analysis ↗ | — | 1 Oct 2026 |
| 45 | Kimi K2.6 | Moonshot AI (Kimi) | 90.5% | Defaultthinking | Maker's own figureVendor-reportedMoonshot AI ↗ | — | 20 Apr 2026 |
| 46 | Muse Spark 1.2 | Meta (Meta Superintelligence Labs, Muse) | 90.4% | Extra highMuse Spark 1.2 (Xhigh) | Independent testIndependentArtificial Analysis ↗ | — | 1 Oct 2026 |
| 47 | Qwen3.6 Plus | Qwen (Alibaba) | 90.4% | Defaultthinking | Maker's own figureVendor-reportedQwen (Alibaba) ↗ | — | 2 Apr 2026 |
| 48 | Qwen3.7 Plus | Qwen (Alibaba) | 90.3% | Defaultthinking | Maker's own figureVendor-reportedQwen (Alibaba) ↗ | — | 21 May 2026 |
| 49 | Qwen3.7 Max | Qwen (Alibaba) | 90.2% | Default | Independent testIndependentVals.ai ↗ | — | 1 Sep 2026 |
| 50 | DeepSeek V4 Pro (Preview, 0423) | DeepSeek | 90.1% | MaxThink Max | Maker's own figureVendor-reportedDeepSeek ↗ | — | 24 Apr 2026 |
| 51 | Muse Spark 1.1 | Meta (Meta Superintelligence Labs, Muse) | 89.8% | Extra highMuse Spark 1.1 (Xhigh) | Independent testIndependentArtificial Analysis ↗ | — | 1 Oct 2026 |
| 52 | Hy3 | Tencent Hunyuan | 89.7% | DefaultHy3 | Independent testIndependentArtificial Analysis ↗ | — | 1 Oct 2026 |
| 53 | DeepSeek V4 Flash (Preview, 0423) | DeepSeek | 88.1% | MaxThink Max | Maker's own figureVendor-reportedDeepSeek ↗ | — | 24 Apr 2026 |
| 54 | Qwen3.6 27B | Qwen (Alibaba) | 87.8% | Defaultthinking | Maker's own figureVendor-reportedQwen (Alibaba) ↗ | — | 22 Apr 2026 |
| 55 | Gemini 3.1 Flash-Lite | Google (Gemini / DeepMind) | 86.9% | High | Maker's own figureVendor-reportedGoogle ↗ | — | 3 Mar 2026 |
| 56 | Nemotron 3 Ultra | NVIDIA | 86.7% | DefaultNemotron 3 Ultra 550B A55B (Reasoning) | Independent testIndependentArtificial Analysis ↗ | — | 1 Oct 2026 |
| 57 | GLM-5.1 | Z.ai (Zhipu) | 86.2% | Defaultthinking | Maker's own figureVendor-reportedZ.ai ↗ | — | 7 Apr 2026 |
| 58 | Gemma 4 31B IT | Google (Gemini / DeepMind) | 85.7% | DefaultGemma 4 31B (Reasoning) | Independent testIndependentArtificial Analysis ↗ | — | 1 Oct 2026 |
| 59 | Qwen3.6 35B A3B | Qwen (Alibaba) | 84.1% | DefaultQwen3.6 35B A3B (Reasoning) | Independent testIndependentArtificial Analysis ↗ | — | 1 Oct 2026 |
| 60 | Muse Glimmer 30B | Meta (Meta Superintelligence Labs, Muse) | 83.5% | HighMuse Glimmer (High) | Independent testIndependentArtificial Analysis ↗ | — | 1 Oct 2026 |
| 61 | Gemma 4 26B A4B IT | Google (Gemini / DeepMind) | 82.3% | DefaultThinking | Maker's own figureVendor-reportedGoogle ↗ | — | 2 Apr 2026 |
| 62 | Nemotron 3 Super 120B A12B | NVIDIA | 80.0% | DefaultNemotron 3 Super 120B A12B (Reasoning) | Independent testIndependentArtificial Analysis ↗ | — | 1 Oct 2026 |
| 63 | gpt-oss-120b | OpenAI | 78.2% | Highgpt-oss-120b (High) | Independent testIndependentArtificial Analysis ↗ | — | 1 Oct 2026 |
| 64 | Mistral Small 4 | Mistral AI | 76.9% | HighMistral Small 4 (Reasoning) | Independent testIndependentArtificial Analysis ↗ | — | 1 Oct 2026 |
| 65 | Command A+ | Cohere | 76.1% | DefaultCommand A+ | Independent testIndependentArtificial Analysis ↗ | — | 1 Oct 2026 |
| 66 | Mistral Medium 3.5 | Mistral AI | 74.8% | DefaultMistral Medium 3.5 | Independent testIndependentArtificial Analysis ↗ | — | 1 Oct 2026 |
| 67 | Nemotron 3.5 Lightning 30B-A3B | NVIDIA | 74.3% | DefaultNemotron 3.5 Lightning | Independent testIndependentArtificial Analysis ↗ | — | 1 Oct 2026 |
| 68 | gpt-oss-20b | OpenAI | 68.8% | Highgpt-oss-20b (High) | Independent testIndependentArtificial Analysis ↗ | — | 1 Oct 2026 |
| 69 | Granite 4.2 8B | IBM Granite | 63.1% | DefaultGranite 4.2 8B | Independent testIndependentArtificial Analysis ↗ | — | 1 Oct 2026 |
| 70 | Ministral 3 14B | Mistral AI | 57.2% | DefaultMinistral 3 14B | Independent testIndependentArtificial Analysis ↗ | — | 1 Oct 2026 |