GPQA Diamond (GPQA-D)
We show this table for reference; we do not rank on it.
Data verified
A display-only GPQA Diamond reference from provider comparison charts.
Benchmark score on GPQA-D — October 29, 2026
We compile the GPQA-D rows from provider self-reports and secondary reports. GPT-6 Astra leads the table at 96.0%, followed by Sakana Fugu-Ultra (95.5%) and Sakana Fugu (95.5%). We do not use these results to rank models overall.
1Closed
GPT-6 Astra
OpenAI
96.0%
Overall 86.07Context 1.05M
2Closed
Sakana Fugu-Ultra
Sakana AI
95.5%
Context 1M
3Closed
Sakana Fugu
Sakana AI
95.5%
Context 1M
86 modelsKnowledgeStaleDisplay onlyUpdated October 29, 2026
Benchmark score table (86 models)
86 of 86 rows
| Rank | Model / configuration | Score | Parameters (B) | Open / closed |
|---|---|---|---|---|
| 1 | GPT-6 AstraOpenAI | 96.0% | Not reported | Closed |
| 2 | Sakana Fugu-UltraSakana AI | 95.5% | Not reported | Closed |
| 3 | Sakana FuguSakana AI | 95.5% | Not reported | Closed |
| 4 | GPT-5.6 SolOpenAI | 94.6% | Not reported | Closed |
| 5 | Gemini 3.1 ProGoogle | 94.3% | Not reported | Closed |
| 6 | Claude Opus 4.7 (Adaptive)Anthropic | 94.2% | Not reported | Closed |
| 7 | Kimi K3.1 ProMoonshot AI | 94.2% | Not reported | Pending |
| 8 | Claude Opus 4.8Anthropic | 93.6% | Not reported | Closed |
| 9 | GPT-5.5OpenAI | 93.6% | Not reported | Closed |
| 10 | Kimi K3Moonshot AI | 93.5% | Not reported | Open |
| 11 | Step 5 PreviewStepFun | 93.5% | Not reported | Open |
| 12 | GPT-5.6 TerraOpenAI | 92.9% | Not reported | Closed |
| 13 | GPT-5.4OpenAI | 92.8% | Not reported | Closed |
| 14 | Ornith-1.5-397BOrnith AI | 92.8% | Not reported | Open |
| 15 | Gemini 3.5 FlashGoogle | 92.7% | Not reported | Closed |
| 16 | Qwen3.8 MaxAlibaba | 92.6% | Not reported | Open |
| 17 | Qwen3.7 MaxAlibaba | 92.4% | Not reported | Closed |
| 18 | Pareto 26.10 PreviewUnbiased | 92.4% | Not reported | Closed |
| 19 | GPT-5.6 LunaOpenAI | 92.3% | Not reported | Closed |
| 20 | Hy4 previewTencent | 92.3% | Not reported | Open |
| 21 | Qwen3.8-Flash-NextAlibaba | 91.7% | Not reported | Open |
| 22 | GLM-5.2Z.AI | 91.2% | Not reported | Open |
| 23 | Qwen3.8-Omni-FlashAlibaba | 91.0% | Not reported | Closed |
| 24 | DeepSeek V4.1 Flash 1017DeepSeek | 90.9% | Not reported | Open |
| 25 | Beam 1.0 501B-A23BReflection AI | 90.6% | Not reported | Open |
| 26 | Kimi K2.6Moonshot AI | 90.5% | Not reported | Open |
| 27 | Particle-1Particle AI | 90.4% | Not reported | Closed |
| 28 | Qwen3.7 PlusAlibaba | 90.3% | Not reported | Closed |
| 29 | Beam 1.0 501B-A23B FP8Reflection AI | 90.3% | Not reported | Open |
| 30 | DeepSeek V4 Pro 0813DeepSeek | 90.1% | Not reported | Open |
| 31 | Interfaze BetaInterfaze | 89.9% | Not reported | Closed |
| 32 | Nemotron 3.5 Super VL 120B A12B NVFP4NVIDIA | 89.9% | Not reported | Open |
| 33 | Nemotron 3.5 Super VL 120B A12BNVIDIA | 89.5% | Not reported | Open |
| 34 | Inkling-SmallThinking Machines Lab | 89.5% | Not reported | Open |
| 35 | Muse SparkMeta | 89.5% | Not reported | Closed |
| 36 | Beam 1.0 501B-A23B NVFP4Reflection AI | 89.3% | Not reported | Open |
| 37 | Mercury 3Inception | 89.2% | Not reported | Closed |
| 38 | Claude Opus 4.6Anthropic | 89.2% | Not reported | Closed |
| 39 | Qwen3.8-27BAlibaba | 89.2% | Not reported | Open |
| 40 | Ornith-1.5-35B-A3BOrnith AI | 89.2% | Not reported | Open |
| 41 | DeepSeek V4 Pro (High)DeepSeek | 89.1% | Not reported | Open |
| 42 | Solar Pro 4Upstage | 89.0% | Not reported | Closed |
| 43 | Nemotron 3.5 Bolt 74B-A9B BF16NVIDIA | 88.8% | Not reported | Open |
| 44 | Grok 4.20xAI | 88.5% | Not reported | Closed |
| 45 | OrcaCyber-Nano-1.0OrcaRouter | 88.3% | Not reported | Open |
| 46 | DeepSeek V4 Flash 0731DeepSeek | 88.1% | Not reported | Open |
| 47 | InklingThinking Machines Lab | 87.9% | Not reported | Open |
| 48 | Kimi K2.5Moonshot AI | 87.6% | Not reported | Open |
| 49 | DeepSeek V4 Flash (High)DeepSeek | 87.4% | Not reported | Open |
| 50 | Hy3 PreviewTencent | 87.2% | Not reported | Open |
| 51 | MiniMax M2.7MiniMax | 87.0% | Not reported | Open |
| 52 | Nemotron 3 UltraNVIDIA | 87.0% | Not reported | Open |
| 53 | Nemotron 3.5 Bolt 74B-A9BNVIDIA | 86.4% | Not reported | Open |
| 54 | Ornith-1.5-9BOrnith AI | 86.4% | Not reported | Open |
| 55 | Solar Open 2Upstage | 86.3% | Not reported | Open |
| 56 | GLM-5.1Z.AI | 86.2% | Not reported | Open |
| 57 | GLM-5Z.AI | 86.0% | Not reported | Open |
| 58 | Ternary Bonsai 2 27BPrism ML | 85.8% | Not reported | Open |
| 59 | A.X K2SK Telecom | 85.6% | Not reported | Open |
| 60 | Ling 3.0 FlashInclusionAI | 85.0% | Not reported | Open |
| 61 | MAI-Thinking-1Microsoft | 84.2% | Not reported | Closed |
| 62 | Ling 3.0 Flash FP8InclusionAI | 84.0% | Not reported | Open |
| 63 | K-EXAONE 2.0LG AI Research | 82.2% | Not reported | Open |
| 64 | Mercury 2.5Inception | 79.0% | Not reported | Closed |
| 65 | Gemma 4 12BGoogle | 78.8% | Not reported | Open |
| 66 | Spark-X2.6-4BSparkLLM | 77.1% | Not reported | Open |
| 67 | Trinity-Large-ThinkingArcee AI | 76.3% | Not reported | Open |
| 68 | Nemotron 3.5 Lightning 30B A3B NVFP4NVIDIA | 75.6% | Not reported | Open |
| 69 | LFM2.5-24B-A2BLiquidAI | 74.4% | Not reported | Open |
| 70 | Mercury 2Inception | 73.6% | Not reported | Closed |
| 71 | DeepSeek V4 ProDeepSeek | 72.9% | Not reported | Open |
| 72 | Nemotron 3 Nano Omni 30B A3BNVIDIA | 72.2% | Not reported | Open |
| 73 | DeepSeek V4 FlashDeepSeek | 71.2% | Not reported | Open |
| 74 | ZAYA1-8BZyphra | 71.0% | Not reported | Open |
| 75 | MiniCPM5-2BOpenBMB | 70.2% | Not reported | Open |
| 76 | LongCat-Flash-Lite-SparseMeituan | 69.5% | Not reported | Open |
| 77 | Mellum2.1-12B-A2.5B-ThinkingJetBrains | 64.6% | Not reported | Open |
| 78 | LFM3-1.7BLiquidAI | 63.4% | Not reported | Open |
| 79 | Trinity-Large-PreviewArcee AI | 63.3% | Not reported | Open |
| 80 | Mellum2-12B-A2.5B-ThinkingJetBrains | 57.6% | Not reported | Open |
| 81 | ZAYA1-74B-PreviewZyphra | 57.3% | Not reported | Open |
| 82 | LLaDA2.2-miniInclusionAI | 44.4% | Not reported | Open |
| 83 | Soofi S 30B-A3BSoofi Project | 43.4% | Not reported | Open |
| 84 | Mellum2-12B-A2.5B-InstructJetBrains | 40.9% | Not reported | Open |
| 85 | MiniCPM5-1BOpenBMB | 26.3% | Not reported | Open |
| 86 | LFM2.5-230MLiquidAI | 25.4% | Not reported | Open |
AA-GPQA Diamond
BenchLM stores the Artificial Analysis GPQA Diamond result separately from the core GPQA lane because Artificial Analysis controls the evaluation configuration.
218 of 218 rows
| Source rank | Source model or configuration | Score | Source setup | Parameters (B) | Open / closed |
|---|---|---|---|---|---|
| 1 | GPT-6 AstraOpenAI | 96.1% | Artificial AnalysisOriginal result | Not reported | Closed |
| 2 | Gemini 3.8 FlashGoogle | 95.3% | Artificial AnalysisOriginal result | Not reported | Closed |
| 3 | Grok 4.6xAI | 94.9% | Artificial AnalysisOriginal result | Not reported | Closed |
| 4 | Gemini 3.7 FlashGoogle | 94.5% | Artificial AnalysisOriginal result | Not reported | Closed |
| 5 | GPT-5.6 SolOpenAI | 94.1% | Artificial AnalysisOriginal result | Not reported | Closed |
| 6 | Gemini 3.1 ProGoogle | 94.1% | Artificial AnalysisOriginal result | Not reported | Closed |
| 7 | Claude Fable 5.1Anthropic | 93.7% | Artificial AnalysisOriginal result | Not reported | Closed |
| 8 | Kimi K3Moonshot AI | 93.5% | Artificial AnalysisOriginal result | Not reported | Open |
| 9 | GPT-5.5OpenAI | 93.5% | Artificial AnalysisOriginal result | Not reported | Closed |
| 10 | Muse Spark 1.3Meta | 93.5% | Artificial AnalysisOriginal result | Not reported | Closed |
| 11 | Claude Opus 5Anthropic | 93.2% | Artificial AnalysisOriginal result | Not reported | Closed |
| 12 | Grok 4.5xAI | 93.1% | Artificial AnalysisOriginal result | Not reported | Closed |
| 13 | MiniMax M3MiniMax | 92.9% | Artificial AnalysisOriginal result | Not reported | Open |
| 14 | DeepSeek V4 Pro 0813DeepSeek | 92.8% | Artificial AnalysisOriginal result | Not reported | Open |
| 15 | Gemini 3.6 FlashGoogle | 92.8% | Artificial AnalysisOriginal result | Not reported | Closed |
| 16 | Qwen3.8 Max PreviewAlibaba | 92.8% | Artificial AnalysisOriginal result | Not reported | Closed |
| 17 | Claude Fable 5Anthropic | 92.6% | Artificial AnalysisOriginal result | Not reported | Closed |
| 18 | GPT-5.6 TerraOpenAI | 92.5% | Artificial AnalysisOriginal result | Not reported | Closed |
| 19 | Qwen3.7 MaxAlibaba | 92.3% | Artificial AnalysisOriginal result | Not reported | Closed |
| 20 | Qwen3.8-Flash-NextAlibaba | 92.3% | Artificial AnalysisOriginal result | Not reported | Open |
| 21 | Gemini 3.5 FlashGoogle | 92.2% | Artificial AnalysisOriginal result | Not reported | Closed |
| 22 | Claude Opus 4.8Anthropic | 92.0% | Artificial AnalysisOriginal result | Not reported | Closed |
| 23 | GPT-5.4OpenAI | 92.0% | Artificial AnalysisOriginal result | Not reported | Closed |
| 24 | GLM-5.3Z.AI | 91.7% | Artificial AnalysisOriginal result | Not reported | Open |
| 25 | GPT-5.3 CodexOpenAI | 91.5% | Artificial AnalysisOriginal result | Not reported | Closed |
| 26 | GPT-5.3-Codex-SparkOpenAI | 91.5% | Artificial AnalysisOriginal result | Not reported | Closed |
| 27 | Claude Opus 4.7 (Adaptive)Anthropic | 91.4% | Artificial AnalysisOriginal result | Not reported | Closed |
| 28 | Claude Sonnet 5Anthropic | 91.1% | Artificial AnalysisOriginal result | Not reported | Closed |
| 29 | GPT-5.6 LunaOpenAI | 91.1% | Artificial AnalysisOriginal result | Not reported | Closed |
| 30 | Kimi K2.6Moonshot AI | 91.1% | Artificial AnalysisOriginal result | Not reported | Open |
| 31 | Gemini 3 ProGoogle | 90.8% | Artificial AnalysisOriginal result | Not reported | Closed |
| 32 | DeepSeek V4 Flash 0731DeepSeek | 90.8% | Artificial AnalysisOriginal result | Not reported | Open |
| 33 | Qwen3.8-27BAlibaba | 90.5% | Artificial AnalysisOriginal result | Not reported | Open |
| 34 | DeepSeek V4 Pro (High)DeepSeek | 90.5% | Artificial AnalysisOriginal result | Not reported | Open |
| 35 | Muse Spark 1.2Meta | 90.4% | Artificial AnalysisOriginal result | Not reported | Closed |
| 36 | GPT-5.2OpenAI | 90.3% | Artificial AnalysisOriginal result | Not reported | Closed |
| 37 | Grok 4.3xAI | 90.1% | Artificial AnalysisOriginal result | Not reported | Closed |
| 38 | Qwen3.7 PlusAlibaba | 90.0% | Artificial AnalysisOriginal result | Not reported | Closed |
| 39 | GPT-5.2-CodexOpenAI | 89.9% | Artificial AnalysisOriginal result | Not reported | Closed |
| 40 | Muse Spark 1.1Meta | 89.8% | Artificial AnalysisOriginal result | Not reported | Closed |
| 41 | Hy3Tencent | 89.7% | Artificial AnalysisOriginal result | Not reported | Open |
| 42 | Hy3 PreviewTencent | 89.7% | Artificial AnalysisOriginal result | Not reported | Open |
| 43 | Kimi K2.7 CodeMoonshot AI | 89.6% | Artificial AnalysisOriginal result | Not reported | Open |
| 44 | Claude Opus 4.6 (Adaptive)Anthropic | 89.6% | Artificial AnalysisOriginal result | Not reported | Closed |
| 45 | GLM-5.2Z.AI | 89.5% | Artificial AnalysisOriginal result | Not reported | Open |
| 46 | Inkling-SmallThinking Machines Lab | 89.5% | Artificial AnalysisOriginal result | Not reported | Open |
| 47 | Qwen3.5 397BAlibaba | 89.3% | Artificial AnalysisOriginal result | Not reported | Open |
| 48 | Qwen3.5 397B (Reasoning)Alibaba | 89.3% | Artificial AnalysisOriginal result | Not reported | Open |
| 49 | Solar Pro 4Upstage | 89.1% | Artificial AnalysisOriginal result | Not reported | Closed |
| 50 | Muse Spark OpenMeta | 89.0% | Artificial AnalysisOriginal result | Not reported | Open |
| 51 | Qwen 3.6 Max (preview)Alibaba | 88.8% | Artificial AnalysisOriginal result | Not reported | Closed |
| 52 | Claude Opus 4.7Anthropic | 88.5% | Artificial AnalysisOriginal result | Not reported | Closed |
| 53 | Muse SparkMeta | 88.4% | Artificial AnalysisOriginal result | Not reported | Closed |
| 54 | Qwen3.6 PlusAlibaba | 88.2% | Artificial AnalysisOriginal result | Not reported | Closed |
| 55 | Kimi K2.5 (Reasoning)Moonshot AI | 87.9% | Artificial AnalysisOriginal result | Not reported | Closed |
| 56 | Kimi K2.5Moonshot AI | 87.9% | Artificial AnalysisOriginal result | Not reported | Open |
| 57 | Grok 4xAI | 87.7% | Artificial AnalysisOriginal result | Not reported | Closed |
| 58 | GPT-5.4 miniOpenAI | 87.5% | Artificial AnalysisOriginal result | Not reported | Closed |
| 59 | MiniMax M2.7MiniMax | 87.4% | Artificial AnalysisOriginal result | Not reported | Open |
| 60 | GPT-5.1OpenAI | 87.3% | Artificial AnalysisOriginal result | Not reported | Closed |
| 61 | InklingThinking Machines Lab | 87.2% | Artificial AnalysisOriginal result | Not reported | Open |
| 62 | MiMo-V2-ProXiaomi | 87.0% | Artificial AnalysisOriginal result | Not reported | Closed |
| 63 | GLM-5.1Z.AI | 86.8% | Artificial AnalysisOriginal result | Not reported | Open |
| 64 | Nemotron 3 UltraNVIDIA | 86.7% | Artificial AnalysisOriginal result | Not reported | Open |
| 65 | MiMo-V2.5-ProXiaomi | 86.6% | Artificial AnalysisOriginal result | Not reported | Closed |
| 66 | Claude Opus 4.5 ThinkingAnthropic | 86.6% | Artificial AnalysisOriginal result | Not reported | Closed |
| 67 | Apodex 1.1Apodex | 86.4% | Artificial AnalysisOriginal result | Not reported | Closed |
| 68 | Apodex 1.1 MiniApodex | 86.4% | Artificial AnalysisOriginal result | Not reported | Open |
| 69 | Ling 3.0 Flash VLInclusionAI | 86.2% | Artificial AnalysisOriginal result | Not reported | Open |
| 70 | GPT-5.1-CodexOpenAI | 86.0% | Artificial AnalysisOriginal result | Not reported | Closed |
| 71 | GPT-5.1-Codex-MaxOpenAI | 86.0% | Artificial AnalysisOriginal result | Not reported | Closed |
| 72 | GLM-4.7Z.AI | 85.9% | Artificial AnalysisOriginal result | Not reported | Open |
| 73 | Qwen3.5-27BAlibaba | 85.8% | Artificial AnalysisOriginal result | Not reported | Open |
| 74 | Qwen3.5-122B-A10BAlibaba | 85.7% | Artificial AnalysisOriginal result | Not reported | Open |
| 75 | Gemma 4 31BGoogle | 85.7% | Artificial AnalysisOriginal result | Not reported | Open |
| 76 | A.X K2SK Telecom | 85.7% | Artificial AnalysisOriginal result | Not reported | Open |
| 77 | Ling 3.0 FlashInclusionAI | 85.5% | Artificial AnalysisOriginal result | Not reported | Open |
| 78 | GPT-5 (high)OpenAI | 85.4% | Artificial AnalysisOriginal result | Not reported | Closed |
| 79 | Grok 4.1 Fast (Reasoning)xAI | 85.3% | Artificial AnalysisOriginal result | Not reported | Closed |
| 80 | MiniMax M2.5MiniMax | 84.8% | Artificial AnalysisOriginal result | Not reported | Closed |
| 81 | GLM-5-TurboZ.AI | 84.7% | Artificial AnalysisOriginal result | Not reported | Closed |
| 82 | Grok 4 Fast (Reasoning)xAI | 84.7% | Artificial AnalysisOriginal result | Not reported | Closed |
| 83 | Qwen3.5-35B-A3BAlibaba | 84.5% | Artificial AnalysisOriginal result | Not reported | Open |
| 84 | o3-proOpenAI | 84.5% | Artificial AnalysisOriginal result | Not reported | Closed |
| 85 | Gemini 2.5 ProGoogle | 84.4% | Artificial AnalysisOriginal result | Not reported | Closed |
| 86 | Qwen3.6-27BAlibaba | 84.2% | Artificial AnalysisOriginal result | Not reported | Open |
| 87 | GPT-5 (medium)OpenAI | 84.2% | Artificial AnalysisOriginal result | Not reported | Closed |
| 88 | Qwen3.6-35B-A3BAlibaba | 84.1% | Artificial AnalysisOriginal result | Not reported | Open |
| 89 | Claude Opus 4.6Anthropic | 84.0% | Artificial AnalysisOriginal result | Not reported | Closed |
| 90 | Gemini 3.5 Flash-LiteGoogle | 83.8% | Artificial AnalysisOriginal result | Not reported | Closed |
| 91 | Muse Glimmer 30BMeta | 83.5% | Artificial AnalysisOriginal result | Not reported | Open |
| 92 | K-EXAONE 2.0LG AI Research | 82.9% | Artificial AnalysisOriginal result | Not reported | Open |
| 93 | MiMo-V2-OmniXiaomi | 82.8% | Artificial AnalysisOriginal result | Not reported | Closed |
| 94 | GPT-5 miniOpenAI | 82.8% | Artificial AnalysisOriginal result | Not reported | Closed |
| 95 | o3OpenAI | 82.7% | Artificial AnalysisOriginal result | Not reported | Closed |
| 96 | Step 3.5 FlashStepFun | 82.6% | Artificial AnalysisOriginal result | Not reported | Open |
| 97 | GLM-5Z.AI | 82.0% | Artificial AnalysisOriginal result | Not reported | Open |
| 98 | GPT-5.4 nanoOpenAI | 81.7% | Artificial AnalysisOriginal result | Not reported | Closed |
| 99 | DeepSeek-R1DeepSeek | 81.3% | Artificial AnalysisOriginal result | Not reported | Open |
| 100 | Gemini 3 FlashGoogle | 81.2% | Artificial AnalysisOriginal result | Not reported | Closed |
| 101 | Claude Opus 4.5Anthropic | 81.0% | Artificial AnalysisOriginal result | Not reported | Closed |
| 102 | Claude 4.1 Opus ThinkingAnthropic | 80.9% | Artificial AnalysisOriginal result | Not reported | Closed |
| 103 | GLM-5V-TurboZ.AI | 80.9% | Artificial AnalysisOriginal result | Not reported | Closed |
| 104 | Step 3.7 FlashStepFun | 80.9% | Artificial AnalysisOriginal result | Not reported | Open |
| 105 | Claude Sonnet 4.6Anthropic | 79.9% | Artificial AnalysisOriginal result | Not reported | Closed |
| 106 | Claude 4 Opus ThinkingAnthropic | 79.6% | Artificial AnalysisOriginal result | Not reported | Closed |
| 107 | Gemma 4 26B A4BGoogle | 79.2% | Artificial AnalysisOriginal result | Not reported | Open |
| 108 | o4-miniOpenAI | 78.4% | Artificial AnalysisOriginal result | Not reported | Closed |
| 109 | K-ExaoneLG AI Research | 78.3% | Artificial AnalysisOriginal result | Not reported | Closed |
| 110 | GPT-OSS 120BOpenAI | 78.2% | Artificial AnalysisOriginal result | Not reported | Open |
| 111 | DeepSeek V3.1 (Reasoning)DeepSeek | 77.9% | Artificial AnalysisOriginal result | Not reported | Open |
| 112 | Claude 3.7 Sonnet ThinkingAnthropic | 77.2% | Artificial AnalysisOriginal result | Not reported | Closed |
| 113 | Qwen3.5-4BAlibaba | 77.1% | Artificial AnalysisOriginal result | Not reported | Open |
| 114 | Mercury 2Inception | 77.0% | Artificial AnalysisOriginal result | Not reported | Closed |
| 115 | Mistral Small 4Mistral | 76.9% | Artificial AnalysisOriginal result | Not reported | Open |
| 116 | Kimi K2Moonshot AI | 76.6% | Artificial AnalysisOriginal result | Not reported | Closed |
| 117 | o1-previewOpenAI | 76.5% | Artificial AnalysisOriginal result | Not reported | Closed |
| 118 | Qwen3 MaxAlibaba | 76.4% | Artificial AnalysisOriginal result | Not reported | Closed |
| 119 | Command A+Cohere | 76.1% | Artificial AnalysisOriginal result | Not reported | Open |
| 120 | North Mini CodeCohere | 75.7% | Artificial AnalysisOriginal result | Not reported | Open |
| 121 | Gemma 4 12BGoogle | 75.3% | Artificial AnalysisOriginal result | Not reported | Open |
| 122 | Trinity-Large-ThinkingArcee AI | 75.2% | Artificial AnalysisOriginal result | Not reported | Open |
| 123 | Trinity-Large-PreviewArcee AI | 75.2% | Artificial AnalysisOriginal result | Not reported | Open |
| 124 | DeepSeek V3.2DeepSeek | 75.1% | Artificial AnalysisOriginal result | Not reported | Open |
| 125 | Mistral Medium 3.5 128BMistral | 74.8% | Artificial AnalysisOriginal result | Not reported | Open |
| 126 | o3-miniOpenAI | 74.8% | Artificial AnalysisOriginal result | Not reported | Closed |
| 127 | o1OpenAI | 74.7% | Artificial AnalysisOriginal result | Not reported | Closed |
| 128 | Nemotron 3.5 Lightning 30B A3B NVFP4NVIDIA | 74.3% | Artificial AnalysisOriginal result | Not reported | Open |
| 129 | Sarvam 105BSarvam | 73.8% | Artificial AnalysisOriginal result | Not reported | Open |
| 130 | DeepSeek V3.1DeepSeek | 73.5% | Artificial AnalysisOriginal result | Not reported | Open |
| 131 | Ling 3.0 TinyInclusionAI | 73.4% | Artificial AnalysisOriginal result | Not reported | Open |
| 132 | GLM-4.5-AirZ.AI | 73.3% | Artificial AnalysisOriginal result | Not reported | Closed |
| 133 | Quasar 438BMultiverse Computing | 73.2% | Artificial AnalysisOriginal result | Not reported | Closed |
| 134 | Nemotron Ultra 253BNVIDIA | 72.8% | Artificial AnalysisOriginal result | Not reported | Open |
| 135 | Grok Code Fast 1xAI | 72.7% | Artificial AnalysisOriginal result | Not reported | Closed |
| 136 | Qwen3-Omni-30B-A3B-ThinkingAlibaba | 72.6% | Artificial AnalysisOriginal result | Not reported | Open |
| 137 | Solar Pro 3Upstage | 72.4% | Artificial AnalysisOriginal result | Not reported | Closed |
| 138 | MiniCPM5-2BOpenBMB | 70.2% | Artificial AnalysisOriginal result | Not reported | Open |
| 139 | Claude 4 OpusAnthropic | 70.1% | Artificial AnalysisOriginal result | Not reported | Closed |
| 140 | MiniMax M1 80kMiniMax | 69.7% | Artificial AnalysisOriginal result | Not reported | Closed |
| 141 | GPT-OSS 20BOpenAI | 68.8% | Artificial AnalysisOriginal result | Not reported | Open |
| 142 | Gemini 2.5 FlashGoogle | 68.3% | Artificial AnalysisOriginal result | Not reported | Closed |
| 143 | Claude 4 SonnetAnthropic | 68.3% | Artificial AnalysisOriginal result | Not reported | Closed |
| 144 | Mistral Large 3Mistral | 68.0% | Artificial AnalysisOriginal result | Not reported | Closed |
| 145 | Magistral MediumMistral | 67.9% | Artificial AnalysisOriginal result | Not reported | Closed |
| 146 | GPT-5 nanoOpenAI | 67.6% | Artificial AnalysisOriginal result | Not reported | Closed |
| 147 | Llama 4 MaverickMeta | 67.1% | Artificial AnalysisOriginal result | Not reported | Open |
| 148 | GPT-4.1OpenAI | 66.6% | Artificial AnalysisOriginal result | Not reported | Closed |
| 149 | GPT-4.1 miniOpenAI | 66.4% | Artificial AnalysisOriginal result | Not reported | Closed |
| 150 | Claude 3.7 SonnetAnthropic | 65.6% | Artificial AnalysisOriginal result | Not reported | Closed |
| 151 | MiMo-V2-FlashXiaomi | 65.6% | Artificial AnalysisOriginal result | Not reported | Open |
| 152 | DeepSeek V3 0324DeepSeek | 65.5% | Artificial AnalysisOriginal result | Not reported | Open |
| 153 | Granite 4.2 30BIBM | 64.4% | Artificial AnalysisOriginal result | Not reported | Open |
| 154 | Magistral SmallMistral | 64.1% | Artificial AnalysisOriginal result | Not reported | Open |
| 155 | Grok 4.1 FastxAI | 63.7% | Artificial AnalysisOriginal result | Not reported | Closed |
| 156 | Sarvam 30BSarvam | 63.3% | Artificial AnalysisOriginal result | Not reported | Open |
| 157 | GLM-4.6Z.AI | 63.2% | Artificial AnalysisOriginal result | Not reported | Open |
| 158 | Celeris-1Celeris | 63.1% | Artificial AnalysisOriginal result | Not reported | Closed |
| 159 | Granite 4.2 8BIBM | 63.1% | Artificial AnalysisOriginal result | Not reported | Open |
| 160 | Exaone 4.0 32BLG AI Research | 62.8% | Artificial AnalysisOriginal result | Not reported | Open |
| 161 | Gemini 2.0 FlashGoogle | 62.3% | Artificial AnalysisOriginal result | Not reported | Closed |
| 162 | Qwen3-Omni-30B-A3B-InstructAlibaba | 62.0% | Artificial AnalysisOriginal result | Not reported | Open |
| 163 | DeepSeek R1 Distill Qwen 32BDeepSeek | 61.5% | Artificial AnalysisOriginal result | Not reported | Open |
| 164 | o1-miniOpenAI | 60.3% | Artificial AnalysisOriginal result | Not reported | Closed |
| 165 | QwQ-32BAlibaba | 59.3% | Artificial AnalysisOriginal result | Not reported | Open |
| 166 | Ling 2.6 FlashInclusionAI | 59.3% | Artificial AnalysisOriginal result | Not reported | Open |
| 167 | Gemini 1.5 ProGoogle | 58.9% | Artificial AnalysisOriginal result | Not reported | Closed |
| 168 | Llama 4 ScoutMeta | 58.7% | Artificial AnalysisOriginal result | Not reported | Open |
| 169 | GLM-4.7-FlashZ.AI | 58.1% | Artificial AnalysisOriginal result | Not reported | Open |
| 170 | Mistral Medium 3Mistral | 57.8% | Artificial AnalysisOriginal result | Not reported | Closed |
| 171 | Gemma 4 E4BGoogle | 57.6% | Artificial AnalysisOriginal result | Not reported | Open |
| 172 | Phi-4Microsoft | 57.5% | Artificial AnalysisOriginal result | Not reported | Open |
| 173 | Phi-4-reasoningMicrosoft | 57.5% | Artificial AnalysisOriginal result | Not reported | Open |
| 174 | Ministral 3 14B (Reasoning)Mistral | 57.2% | Artificial AnalysisOriginal result | Not reported | Open |
| 175 | Ministral 3 14BMistral | 57.2% | Artificial AnalysisOriginal result | Not reported | Open |
| 176 | Mistral Small 4 (Reasoning)Mistral | 57.1% | Artificial AnalysisOriginal result | Not reported | Open |
| 177 | Solar Pro 2Upstage | 56.1% | Artificial AnalysisOriginal result | Not reported | Closed |
| 178 | Granite 4.2 3BIBM | 55.9% | Artificial AnalysisOriginal result | Not reported | Open |
| 179 | LFM2.5-2.6BLiquidAI | 55.8% | Artificial AnalysisOriginal result | Not reported | Open |
| 180 | DeepSeek V3DeepSeek | 55.7% | Artificial AnalysisOriginal result | Not reported | Open |
| 181 | GPT-4oOpenAI | 54.3% | Artificial AnalysisOriginal result | Not reported | Closed |
| 182 | Llama 3.1 405BMeta | 51.5% | Artificial AnalysisOriginal result | Not reported | Open |
| 183 | LFM2.5-8B-A1BLiquidAI | 51.3% | Artificial AnalysisOriginal result | Not reported | Open |
| 184 | GPT-4.1 nanoOpenAI | 51.2% | Artificial AnalysisOriginal result | Not reported | Closed |
| 185 | Nova ProAmazon | 49.9% | Artificial AnalysisOriginal result | Not reported | Closed |
| 186 | Llama 3.3 70BMeta | 49.8% | Artificial AnalysisOriginal result | Not reported | Open |
| 187 | Ultravox v0.6 Llama 3.3 70BFixie AI | 49.8% | Artificial AnalysisOriginal result | Not reported | Open |
| 188 | Llama 3.1 70BMeta | 49.1% | Artificial AnalysisOriginal result | Not reported | Open |
| 189 | Claude 3 OpusAnthropic | 48.9% | Artificial AnalysisOriginal result | Not reported | Closed |
| 190 | Mistral Large 2Mistral | 48.6% | Artificial AnalysisOriginal result | Not reported | Closed |
| 191 | Gemini 2.5 Flash-LiteGoogle | 47.4% | Artificial AnalysisOriginal result | Not reported | Closed |
| 192 | LFM2-24B-A2BLiquidAI | 47.4% | Artificial AnalysisOriginal result | Not reported | Closed |
| 193 | Ministral 3 8B (Reasoning)Mistral | 47.1% | Artificial AnalysisOriginal result | Not reported | Open |
| 194 | Ministral 3 8BMistral | 47.1% | Artificial AnalysisOriginal result | Not reported | Open |
| 195 | Nemotron 3 Nano Omni 30B A3BNVIDIA | 46.9% | Artificial AnalysisOriginal result | Not reported | Open |
| 196 | Gemini 1.5 FlashGoogle | 46.3% | Artificial AnalysisOriginal result | Not reported | Closed |
| 197 | Mistral Small 3.1Mistral | 45.4% | Artificial AnalysisOriginal result | Not reported | Open |
| 198 | Gemma 4 E2BGoogle | 43.3% | Artificial AnalysisOriginal result | Not reported | Open |
| 199 | Gemma 3 27BGoogle | 42.8% | Artificial AnalysisOriginal result | Not reported | Open |
| 200 | GPT-4o miniOpenAI | 42.6% | Artificial AnalysisOriginal result | Not reported | Closed |
| 201 | Exaone 4.0 1.2BLG AI Research | 42.4% | Artificial AnalysisOriginal result | Not reported | Open |
| 202 | DeepSeek-V2.5DeepSeek | 42.3% | Artificial AnalysisOriginal result | Not reported | Open |
| 203 | Qwen2.5 Coder 32B InstructAlibaba | 41.7% | Artificial AnalysisOriginal result | Not reported | Open |
| 204 | Devstral SmallMistral | 41.4% | Artificial AnalysisOriginal result | Not reported | Open |
| 205 | Claude 3.5 HaikuAnthropic | 40.8% | Artificial AnalysisOriginal result | Not reported | Closed |
| 206 | Nemotron 3 Nano 30BNVIDIA | 39.9% | Artificial AnalysisOriginal result | Not reported | Open |
| 207 | Claude 3 HaikuAnthropic | 37.4% | Artificial AnalysisOriginal result | Not reported | Closed |
| 208 | Ministral 3 3B (Reasoning)Mistral | 35.8% | Artificial AnalysisOriginal result | Not reported | Open |
| 209 | Ministral 3 3BMistral | 35.8% | Artificial AnalysisOriginal result | Not reported | Open |
| 210 | LFM2.5-1.2B-ThinkingLiquidAI | 33.9% | Artificial AnalysisOriginal result | Not reported | Closed |
| 211 | LFM2.5-1.2B-InstructLiquidAI | 32.6% | Artificial AnalysisOriginal result | Not reported | Closed |
| 212 | Phi-4 Multimodal InstructMicrosoft | 31.5% | Artificial AnalysisOriginal result | Not reported | Open |
| 213 | LFM2.5-VL-1.6B-ExtractLiquidAI | 28.9% | Artificial AnalysisOriginal result | Not reported | Open |
| 214 | Gemini 1.0 ProGoogle | 27.7% | Artificial AnalysisOriginal result | Not reported | Closed |
| 215 | Llama 3.1 8BMeta | 27.0% | Artificial AnalysisOriginal result | Not reported | Open |
| 216 | Granite-4.0-H-1BIBM | 26.3% | Artificial AnalysisOriginal result | Not reported | Open |
| 217 | Granite-4.0-350MIBM | 26.1% | Artificial AnalysisOriginal result | Not reported | Open |
| 218 | Granite-4.0-H-350MIBM | 25.7% | Artificial AnalysisOriginal result | Not reported | Open |
GPQA Diamond (Vals)
BenchLM mirrors this Vals AI leaderboard for reference only. It is excluded from BenchLM overall and category rankings.
138 of 138 rows
| Source rank | Source model or configuration | Vals GPQA Diamond mirror score | Source setup | Parameters (B) | Open / closed |
|---|---|---|---|---|---|
| 1 | Gemini 3.1 Pro PreviewGoogle | 95.5% | high reasoningOriginal result | Not reported | Unknown |
| 2 | GPT-5.6 SolOpenAI | 95.2% | max reasoningOriginal result | Not reported | Closed |
| 3 | Grok 4.6xAI | 94.7% | high reasoningOriginal result | Not reported | Closed |
| 4 | Gemini 3.8 FlashGoogle | 94.4% | high reasoningOriginal result | Not reported | Closed |
| 5 | Gemini 3.7 FlashGoogle | 93.9% | high reasoningOriginal result | Not reported | Closed |
| 6 | Qwen3.8 MaxAlibaba | 93.7% | Vals AIOriginal result | Not reported | Open |
| 7 | Gemini 3.6 FlashGoogle | 93.4% | high reasoningOriginal result | Not reported | Closed |
| 8 | Claude Opus 5Anthropic | 93.4% | Vals AIOriginal result | Not reported | Closed |
| 9 | Claude Fable 5.1Anthropic | 93.4% | Vals AIOriginal result | Not reported | Closed |
| 10 | Claude Fable 5Anthropic | 93.2% | Vals AIOriginal result | Not reported | Closed |
| 11 | GPT-5.5OpenAI | 93.2% | xhigh reasoningOriginal result | Not reported | Closed |
| 12 | Kimi K3Moonshot AI | 92.9% | Vals AIOriginal result | Not reported | Open |
| 13 | Grok 4.5xAI | 92.9% | high reasoningOriginal result | Not reported | Closed |
| 14 | Gemini 3.5 FlashGoogle | 92.7% | high reasoningOriginal result | Not reported | Closed |
| 15 | MiniMax M3MiniMax | 92.7% | Vals AIOriginal result | Not reported | Open |
| 16 | Claude Opus 4.8Anthropic | 92.4% | Vals AIOriginal result | Not reported | Closed |
| 17 | DeepSeek V4 Pro 0813DeepSeek | 92.4% | max reasoningOriginal result | Not reported | Open |
| 18 | GPT-5.6 LunaOpenAI | 91.7% | max reasoningOriginal result | Not reported | Closed |
| 19 | Gemini 3 Pro PreviewGoogle | 91.7% | high reasoningOriginal result | Not reported | Unknown |
| 20 | GPT-5.2OpenAI | 91.7% | xhigh reasoningOriginal result | Not reported | Closed |
| 21 | GPT-5.4OpenAI | 91.7% | xhigh reasoningOriginal result | Not reported | Closed |
| 22 | Grok 4.3xAI | 91.4% | Vals AIOriginal result | Not reported | Closed |
| 23 | Muse Spark 1.1Meta | 91.2% | xhigh reasoningOriginal result | Not reported | Closed |
| 24 | GPT-5.6 TerraOpenAI | 90.9% | xhigh reasoningOriginal result | Not reported | Closed |
| 25 | Qwen3.7 MaxAlibaba | 90.2% | Vals AIOriginal result | Not reported | Closed |
| 26 | Claude Opus 4.7Anthropic | 90.2% | Vals AIOriginal result | Not reported | Closed |
| 27 | DeepSeek V4 Flash 0731DeepSeek | 89.9% | high reasoningOriginal result | Not reported | Open |
| 28 | Muse SparkMeta | 89.6% | Vals AIOriginal result | Not reported | Closed |
| 29 | Claude Opus 4.6 (Adaptive)Anthropic | 89.6% | Vals AIOriginal result | Not reported | Closed |
| 30 | DeepSeek V4 Pro 0813DeepSeek | 89.4% | max reasoningOriginal result | Not reported | Open |
| 31 | Kimi K2.6Moonshot AI | 89.1% | Vals AIOriginal result | Not reported | Open |
| 32 | Claude Sonnet 5Anthropic | 88.9% | Vals AIOriginal result | Not reported | Closed |
| 33 | Qwen3.8-27BAlibaba | 88.9% | xhigh reasoningOriginal result | Not reported | Open |
| 34 | Grok 4.20 0309 ReasoningxAI | 88.6% | Vals AIOriginal result | Not reported | Unknown |
| 35 | Grok 4 0709xAI | 88.1% | Vals AIOriginal result | Not reported | Unknown |
| 36 | GLM-5.3Z.AI | 88.1% | max reasoningOriginal result | Not reported | Open |
| 37 | Gemini 3 Flash PreviewGoogle | 87.9% | high reasoningOriginal result | Not reported | Unknown |
| 38 | Qwen3.5 Plus ThinkingAlibaba | 87.4% | Vals AIOriginal result | Not reported | Unknown |
| 39 | Qwen3.6 PlusAlibaba | 87.4% | Vals AIOriginal result | Not reported | Closed |
| 40 | InklingThinking Machines Lab | 87.1% | 0.99 reasoningOriginal result | Not reported | Open |
| 41 | GPT-5.1OpenAI | 86.6% | high reasoningOriginal result | Not reported | Closed |
| 42 | MiniMax M2.7MiniMax | 86.6% | Vals AIOriginal result | Not reported | Open |
| 43 | GLM-5.3-FlashZ.AI | 86.4% | max reasoningOriginal result | Not reported | Open |
| 44 | Nemotron 3 Ultra 550b A55bNvidia | 86.1% | Vals AIOriginal result | Not reported | Unknown |
| 45 | Claude Opus 4.5 ThinkingAnthropic | 85.9% | Vals AIOriginal result | Not reported | Closed |
| 46 | GPT-5OpenAI | 85.6% | high reasoningOriginal result | Not reported | Unknown |
| 47 | GLM-5.2Z.AI | 85.6% | Vals AIOriginal result | Not reported | Open |
| 48 | Claude Sonnet 4.6Anthropic | 85.6% | Vals AIOriginal result | Not reported | Closed |
| 49 | Grok 4 Fast (Reasoning)xAI | 85.4% | Vals AIOriginal result | Not reported | Closed |
| 50 | Ling 3.0 Flash 2607Ant | 84.8% | Vals AIOriginal result | Not reported | Unknown |
| 51 | Qwen3 MaxAlibaba | 84.8% | Vals AIOriginal result | Not reported | Closed |
| 52 | GLM-5.1Z.AI | 84.5% | Vals AIOriginal result | Not reported | Open |
| 53 | Grok 4.1 Fast (Reasoning)xAI | 84.3% | Vals AIOriginal result | Not reported | Closed |
| 54 | o3OpenAI | 84.1% | high reasoningOriginal result | Not reported | Closed |
| 55 | Kimi K2.5 ThinkingMoonshot AI | 84.1% | Vals AIOriginal result | Not reported | Unknown |
| 56 | Gemini 3.5 Flash-LiteGoogle | 83.8% | high reasoningOriginal result | Not reported | Closed |
| 57 | Inkling-SmallThinking Machines Lab | 83.6% | 0.99 reasoningOriginal result | Not reported | Open |
| 58 | GLM 5 ThinkingZhipu AI | 83.3% | Vals AIOriginal result | Not reported | Unknown |
| 59 | GPT-5.4 miniOpenAI | 83.1% | xhigh reasoningOriginal result | Not reported | Closed |
| 60 | Qwen3.5 FlashAlibaba | 82.8% | Vals AIOriginal result | Not reported | Closed |
| 61 | MiMo-V2.5-ProXiaomi | 82.6% | Vals AIOriginal result | Not reported | Closed |
| 62 | MiniMax M2.5MiniMax | 82.1% | Vals AIOriginal result | Not reported | Closed |
| 63 | Claude Sonnet 4.5 ThinkingAnthropic | 81.6% | Vals AIOriginal result | Not reported | Closed |
| 64 | Gemini 2.5 Flash Preview 09 2025Google | 81.6% | Vals AIOriginal result | Not reported | Unknown |
| 65 | MiMo-V2.5Xiaomi | 81.6% | Vals AIOriginal result | Not reported | Closed |
| 66 | Gemini 3.1 Flash Lite PreviewGoogle | 81.1% | high reasoningOriginal result | Not reported | Unknown |
| 67 | Gemini 2.5 Pro Exp 03 25Google | 80.8% | Vals AIOriginal result | Not reported | Unknown |
| 68 | GPT-5 miniOpenAI | 80.3% | high reasoningOriginal result | Not reported | Closed |
| 69 | DeepSeek V3p2 ThinkingFireworks AI | 80.3% | high reasoningOriginal result | Not reported | Unknown |
| 70 | GLM-4.7Z.AI | 80.0% | Vals AIOriginal result | Not reported | Open |
| 71 | Claude Opus 4.5Anthropic | 79.5% | Vals AIOriginal result | Not reported | Closed |
| 72 | Qwen3 MaxAlibaba | 79.5% | Vals AIOriginal result | Not reported | Closed |
| 73 | Grok 3 Mini Fast High ReasoningxAI | 79.3% | high reasoningOriginal result | Not reported | Unknown |
| 74 | GPT-OSS 120BOpenAI | 78.5% | Vals AIOriginal result | Not reported | Open |
| 75 | MiniMax M2.1MiniMax | 78.5% | Vals AIOriginal result | Not reported | Unknown |
| 76 | Kimi K2 ThinkingMoonshot AI | 78.5% | Vals AIOriginal result | Not reported | Unknown |
| 77 | Qwen3 Max PreviewAlibaba | 77.8% | Vals AIOriginal result | Not reported | Unknown |
| 78 | GPT-5.4 nanoOpenAI | 77.5% | high reasoningOriginal result | Not reported | Closed |
| 79 | Gemini 2.5 Flash Preview 09 2025 ThinkingGoogle | 76.5% | Vals AIOriginal result | Not reported | Unknown |
| 80 | Claude Opus 4.1 20250805 ThinkingAnthropic | 76.3% | Vals AIOriginal result | Not reported | Unknown |
| 81 | DeepSeek V3p2Fireworks AI | 76.3% | none reasoningOriginal result | Not reported | Unknown |
| 82 | o3-miniOpenAI | 75.5% | high reasoningOriginal result | Not reported | Closed |
| 83 | Claude 3.7 Sonnet ThinkingAnthropic | 75.3% | Vals AIOriginal result | Not reported | Closed |
| 84 | Claude Sonnet 4 20250514 ThinkingAnthropic | 75.0% | Vals AIOriginal result | Not reported | Unknown |
| 85 | o4-miniOpenAI | 74.5% | high reasoningOriginal result | Not reported | Closed |
| 86 | GLM-4.6Z.AI | 74.5% | Vals AIOriginal result | Not reported | Open |
| 87 | Grok 3xAI | 74.2% | Vals AIOriginal result | Not reported | Unknown |
| 88 | o1OpenAI | 73.2% | high reasoningOriginal result | Not reported | Closed |
| 89 | Grok 3 Mini Fast Low ReasoningxAI | 73.0% | low reasoningOriginal result | Not reported | Unknown |
| 90 | Claude Haiku 4.5 ThinkingAnthropic | 72.2% | Vals AIOriginal result | Not reported | Closed |
| 91 | GLM-4.5Z.AI | 72.2% | Vals AIOriginal result | Not reported | Closed |
| 92 | Claude Opus 4Anthropic | 71.7% | Vals AIOriginal result | Not reported | Unknown |
| 93 | Moonshotai Kimi K2 InstructTogether AI | 71.5% | Vals AIOriginal result | Not reported | Unknown |
| 94 | Gemini 2.5 Flash Lite Preview 09 2025 ThinkingGoogle | 70.2% | Vals AIOriginal result | Not reported | Unknown |
| 95 | Qwen3 235B-A22BAlibaba | 70.2% | Vals AIOriginal result | Not reported | Open |
| 96 | Claude Opus 4.1Anthropic | 70.0% | Vals AIOriginal result | Not reported | Unknown |
| 97 | Llama4 Maverick Instruct BasicFireworks AI | 69.4% | Vals AIOriginal result | Not reported | Unknown |
| 98 | Claude Sonnet 4Anthropic | 69.4% | Vals AIOriginal result | Not reported | Unknown |
| 99 | GPT-OSS 20BOpenAI | 68.9% | Vals AIOriginal result | Not reported | Open |
| 100 | Mistral Large 2512Mistral AI | 68.4% | Vals AIOriginal result | Not reported | Unknown |
| 101 | GPT-4.1 miniOpenAI | 67.9% | high reasoningOriginal result | Not reported | Closed |
| 102 | Claude 3.7 SonnetAnthropic | 67.4% | Vals AIOriginal result | Not reported | Closed |
| 103 | GPT-4.1OpenAI | 65.4% | high reasoningOriginal result | Not reported | Closed |
| 104 | Gemini 2.0 Flash 001Google | 65.2% | Vals AIOriginal result | Not reported | Unknown |
| 105 | Grok 4.1 Fast Non ReasoningxAI | 65.2% | Vals AIOriginal result | Not reported | Unknown |
| 106 | Gemini 2.5 Flash Lite Preview 09 2025Google | 64.1% | Vals AIOriginal result | Not reported | Unknown |
| 107 | GPT-5 nanoOpenAI | 63.4% | high reasoningOriginal result | Not reported | Closed |
| 108 | Langston Nim Nvidia Llama 3.3 Nemotron Super 49b V1 42e84561 ThinkingTogether AI | 62.4% | Vals AIOriginal result | Not reported | Unknown |
| 109 | Magistral Medium 2509Mistral AI | 62.4% | Vals AIOriginal result | Not reported | Unknown |
| 110 | Grok 4 Fast Non ReasoningxAI | 62.1% | Vals AIOriginal result | Not reported | Unknown |
| 111 | DeepSeek V3 0324DeepSeek | 61.6% | Vals AIOriginal result | Not reported | Open |
| 112 | Claude 3.5 SonnetAnthropic | 59.3% | Vals AIOriginal result | Not reported | Closed |
| 113 | MiMo-V2-FlashXiaomi | 59.3% | Vals AIOriginal result | Not reported | Open |
| 114 | Gemini 1.5 Pro 002Google | 58.3% | Vals AIOriginal result | Not reported | Unknown |
| 115 | Magistral Small 2509Mistral AI | 58.3% | Vals AIOriginal result | Not reported | Unknown |
| 116 | Gemini 2.5 Flash Preview 04 17Google | 57.6% | Vals AIOriginal result | Not reported | Unknown |
| 117 | Laguna XS.2Poolside | 55.0% | Vals AIOriginal result | Not reported | Open |
| 118 | DeepSeek V3DeepSeek | 54.5% | Vals AIOriginal result | Not reported | Open |
| 119 | GPT-4oOpenAI | 53.8% | high reasoningOriginal result | Not reported | Closed |
| 120 | GPT-4.1 nanoOpenAI | 50.8% | high reasoningOriginal result | Not reported | Closed |
| 121 | Mistral Small 2402Mistral AI | 50.8% | Vals AIOriginal result | Not reported | Unknown |
| 122 | Grok 2 1212xAI | 50.8% | Vals AIOriginal result | Not reported | Unknown |
| 123 | GPT-4oOpenAI | 50.3% | high reasoningOriginal result | Not reported | Closed |
| 124 | Meta Llama Llama 3.3 70B Instruct TurboTogether AI | 50.0% | Vals AIOriginal result | Not reported | Unknown |
| 125 | Command A 03 2025Cohere | 48.5% | Vals AIOriginal result | Not reported | Unknown |
| 126 | Mistral Large 2411Mistral AI | 47.7% | Vals AIOriginal result | Not reported | Unknown |
| 127 | Langston Nim Nvidia Llama 3.3 Nemotron Super 49b V1 42e84561Together AI | 47.7% | Vals AIOriginal result | Not reported | Unknown |
| 128 | Meta Llama Llama 4 Scout 17B 16E InstructTogether AI | 47.0% | Vals AIOriginal result | Not reported | Unknown |
| 129 | Gemini 1.5 Flash 002Google | 46.0% | Vals AIOriginal result | Not reported | Unknown |
| 130 | Mistral Small 2503Mistral AI | 44.2% | Vals AIOriginal result | Not reported | Unknown |
| 131 | GPT-4o miniOpenAI | 44.2% | high reasoningOriginal result | Not reported | Closed |
| 132 | Claude 3.5 HaikuAnthropic | 37.9% | Vals AIOriginal result | Not reported | Closed |
| 133 | Jamba Large 1.6AI21 Labs | 36.1% | Vals AIOriginal result | Not reported | Unknown |
| 134 | Mistral Medium 3.5Mistral AI | 34.8% | high reasoningOriginal result | Not reported | Unknown |
| 135 | Jamba Mini 1.6AI21 Labs | 32.1% | Vals AIOriginal result | Not reported | Unknown |
| 136 | Command R PlusCohere | 31.1% | Vals AIOriginal result | Not reported | Unknown |
| 137 | GPT-3.5 TurboOpenAI | 30.6% | high reasoningOriginal result | Not reported | Unknown |
| 138 | Laguna M.1Poolside | 27.0% | Vals AIOriginal result | Not reported | Closed |
GPQA Diamond Zero-Shot (Kaggle)
Native proportions are displayed as percentages. Source profiles, task versions, evaluation dates, and symmetric intervals stay attached to each row. Missing results are not zero. How Kaggle Benchmarks reruns tests.
18 of 18 rows
| Source rank | Source model or configuration | Accuracy (%) | Source setup | Parameters (B) | Open / closed |
|---|---|---|---|---|---|
| 1 | Gemini 3.6 Flash | 93.4% | Task version 5; Evaluated 2026-10-08Original result | Not reported | Unknown |
| 2 | GPT-5.6 Sol | 92.9% | Task version 5; Evaluated 2026-10-08Original result | Not reported | Unknown |
| 3 | GPT-6 Astra | 92.9% | Task version 5; Evaluated 2026-10-08Original result | Not reported | Unknown |
| 4 | Gemini 3.5 Flash | 91.4% | Task version 5; Evaluated 2026-10-08Original result | Not reported | Unknown |
| 5 | Grok 4.5 | 90.9% | Task version 5; Evaluated 2026-10-09Original result | Not reported | Unknown |
| 6 | Claude Opus 5 | 90.9% | Task version 5; Evaluated 2026-10-09Original result | Not reported | Unknown |
| 7 | Gemini 3 Flash Preview | 90.4% | Task version 5; Evaluated 2026-10-08Original result | Not reported | Unknown |
| 8 | GPT-5.6 Terra | 86.4% | Task version 5; Evaluated 2026-10-08Original result | Not reported | Unknown |
| 9 | Gemma 4 31B | 84.8% | Task version 5; Evaluated 2026-10-07Original result | Not reported | Unknown |
| 10 | GPT-5 | 84.3% | Task version 5; Evaluated 2026-10-12Original result | Not reported | Unknown |
| 11 | GPT-5.6 Luna | 84.3% | Task version 5; Evaluated 2026-10-09Original result | Not reported | Unknown |
| 12 | Claude Opus 4.8 | 84.3% | Task version 5; Evaluated 2026-10-09Original result | Not reported | Unknown |
| 13 | Gemma 4 26B A4B | 83.2% | Task version 5; Evaluated 2026-10-09Original result | Not reported | Unknown |
| 14 | DeepSeek-R1 | 80.8% | Task version 5; Evaluated 2026-10-09Original result | Not reported | Unknown |
| 15 | Gemini 3.5 Flash-Lite | 74.7% | Task version 5; Evaluated 2026-10-07Original result | Not reported | Unknown |
| 16 | o4 mini | 71.7% | Task version 5; Evaluated 2026-10-12Original result | Not reported | Unknown |
| 17 | gpt-oss-120b | 70.7% | Task version 5; Evaluated 2026-10-09Original result | Not reported | Unknown |
| 18 | gpt-oss-20b | 62.1% | Task version 5; Evaluated 2026-10-12Original result | Not reported | Unknown |
GPQA Diamond Few-Shot (Kaggle)
Native proportions are displayed as percentages. Source profiles, task versions, evaluation dates, and symmetric intervals stay attached to each row. Missing results are not zero. How Kaggle Benchmarks reruns tests.
55 of 55 rows
| Source rank | Source model or configuration | Accuracy (%) | Source setup | Parameters (B) | Open / closed |
|---|---|---|---|---|---|
| 1 | GPT-6 Astra | 93.9% | Task version 4; Evaluated 2026-09-15Original result | Not reported | Unknown |
| 2 | Grok 4.5 | 93.9% | Task version 4; Evaluated 2026-07-14Original result | Not reported | Unknown |
| 3 | Gemini 3 Pro Preview | 93.4% | Task version 4; Evaluated 2025-11-13Original result | Not reported | Unknown |
| 4 | Gemini 3.5 Flash | 92.4% | Task version 4; Evaluated 2026-06-28Original result | Not reported | Unknown |
| 5 | GPT-5.6 Sol | 92.4% | Task version 4; Evaluated 2026-07-15Original result | Not reported | Unknown |
| 6 | Claude Opus 5 | 89.9% | Task version 4; Evaluated 2026-07-27Original result | Not reported | Unknown |
| 7 | Gemini 3 Flash Preview | 89.9% | Task version 4; Evaluated 2026-06-28Original result | Not reported | Unknown |
| 8 | Grok 4 | 86.9% | Task version 4; Evaluated 2025-09-03Original result | Not reported | Unknown |
| 9 | Gemini 3.6 Flash | 86.4% | Task version 4; Evaluated 2026-07-22Original result | Not reported | Unknown |
| 10 | Gemma 4 31B | 86.4% | Task version 4; Evaluation date not reportedOriginal result | Not reported | Unknown |
| 11 | GPT-5.6 Terra | 84.3% | Task version 4; Evaluated 2026-07-15Original result | Not reported | Unknown |
| 12 | Gemini 2.5 Pro Preview | 83.8% | Task version 4; Evaluated 2025-06-06Original result | Not reported | Unknown |
| 13 | GPT-5 | 83.3% | Task version 4; Evaluated 2025-08-19Original result | Not reported | Unknown |
| 14 | Claude Opus 4.8 | 83.2% | Task version 4; Evaluated 2026-06-28Original result | Not reported | Unknown |
| 15 | GPT-5.6 Luna | 81.8% | Task version 4; Evaluated 2026-07-15Original result | Not reported | Unknown |
| 16 | DeepSeek-R1 | 80.3% | Task version 4; Evaluated 2025-06-04Original result | Not reported | Unknown |
| 17 | o3 | 80.3% | Task version 4; Evaluated 2025-05-08Original result | Not reported | Unknown |
| 18 | Gemma 4 26B A4B | 79.2% | Task version 4; Evaluation date not reportedOriginal result | Not reported | Unknown |
| 19 | o4 mini | 76.3% | Task version 4; Evaluated 2025-05-08Original result | Not reported | Unknown |
| 20 | o1 | 74.2% | Task version 4; Evaluated 2025-05-08Original result | Not reported | Unknown |
| 21 | o3 mini | 74.2% | Task version 4; Evaluated 2025-05-08Original result | Not reported | Unknown |
| 22 | Claude Opus 4.1 | 73.7% | Task version 4; Evaluated 2025-08-20Original result | Not reported | Unknown |
| 23 | Claude Opus 4 | 72.7% | Task version 4; Evaluated 2025-05-23Original result | Not reported | Unknown |
| 24 | Gemini 3.5 Flash-Lite | 72.7% | Task version 4; Evaluated 2026-07-22Original result | Not reported | Unknown |
| 25 | Grok 3 Mini | 72.2% | Task version 4; Evaluated 2025-05-26Original result | Not reported | Unknown |
| 26 | deepseek-v3.1 | 71.2% | Task version 4; Evaluated 2025-08-26Original result | Not reported | Unknown |
| 27 | Grok 3 | 70.2% | Task version 4; Evaluated 2025-05-26Original result | Not reported | Unknown |
| 28 | Claude 3.7 Sonnet | 66.7% | Task version 4; Evaluated 2025-05-08Original result | Not reported | Unknown |
| 29 | gpt-oss-120b | 66.2% | Task version 4; Evaluated 2025-08-19Original result | Not reported | Unknown |
| 30 | Claude Sonnet 4 | 65.7% | Task version 4; Evaluated 2025-05-23Original result | Not reported | Unknown |
| 31 | Gemini 2.0 Flash | 65.2% | Task version 4; Evaluated 2025-05-08Original result | Not reported | Unknown |
| 32 | GPT-4.1 | 64.6% | Task version 4; Evaluated 2025-05-08Original result | Not reported | Unknown |
| 33 | DeepSeek-V3 | 62.1% | Task version 4; Evaluated 2025-05-08Original result | Not reported | Unknown |
| 34 | Claude 3.5 Sonnet | 61.6% | Task version 4; Evaluated 2025-05-08Original result | Not reported | Unknown |
| 35 | Gemini 1.5 Pro | 60.6% | Task version 4; Evaluated 2025-05-08Original result | Not reported | Unknown |
| 36 | o1 mini | 60.6% | Task version 4; Evaluated 2025-05-08Original result | Not reported | Unknown |
| 37 | Mistral Large 2 | 55.6% | Task version 4; Evaluated 2025-05-08Original result | Not reported | Unknown |
| 38 | Grok 2 | 53.0% | Task version 4; Evaluated 2025-05-08Original result | Not reported | Unknown |
| 39 | GPT-4o | 50.0% | Task version 4; Evaluated 2025-05-08Original result | Not reported | Unknown |
| 40 | Gemini 1.5 Flash | 46.0% | Task version 4; Evaluated 2025-05-08Original result | Not reported | Unknown |
| 41 | Gemma 3 27B | 45.5% | Task version 4; Evaluated 2025-05-08Original result | Not reported | Unknown |
| 42 | GPT-4o mini | 42.9% | Task version 4; Evaluated 2025-05-08Original result | Not reported | Unknown |
| 43 | Claude 3.5 Haiku | 40.9% | Task version 4; Evaluated 2025-05-08Original result | Not reported | Unknown |
| 44 | Granite 4.0 Small | 40.9% | Task version 4; Evaluated 2025-09-30Original result | Not reported | Unknown |
| 45 | gpt-oss-20b | 37.9% | Task version 4; Evaluated 2025-08-19Original result | Not reported | Unknown |
| 46 | Gemma 3 12B | 35.4% | Task version 4; Evaluated 2025-05-08Original result | Not reported | Unknown |
| 47 | Ministral 8B | 34.8% | Task version 4; Evaluated 2025-05-08Original result | Not reported | Unknown |
| 48 | Mixtral 8x22B | 34.8% | Task version 4; Evaluated 2025-05-08Original result | Not reported | Unknown |
| 49 | Codestral | 33.3% | Task version 4; Evaluated 2025-05-08Original result | Not reported | Unknown |
| 50 | Gemini 1.5 Flash 8B | 32.3% | Task version 4; Evaluated 2025-05-08Original result | Not reported | Unknown |
| 51 | GPT-3.5 Turbo | 30.3% | Task version 4; Evaluated 2025-05-08Original result | Not reported | Unknown |
| 52 | Ministral 3B | 30.3% | Task version 4; Evaluated 2025-05-08Original result | Not reported | Unknown |
| 53 | Gemma 3 4B | 26.8% | Task version 4; Evaluated 2025-05-08Original result | Not reported | Unknown |
| 54 | Gemini 2.5 Flash Preview | 25.3% | Task version 4; Evaluated 2025-05-23Original result | Not reported | Unknown |
| 55 | Qwen 3 235B A22B Thinking | 12.1% | Task version 4; Evaluated 2025-08-19Original result | Not reported | Unknown |
Among the reported GPQA-D rows, GPT-6 Astra is first at 96.0%. The third row is 0.5 points behind. The broader top-10 range is 2.5 points, so many of the published results sit in a relatively narrow band.
86 models have been evaluated on GPQA-D. The benchmark falls in the Knowledge category. GPQA-D is currently displayed for reference but excluded from the scoring formula, so it does not directly affect overall rankings.
About GPQA-D
Year
2023
Tasks
Graduate-level science questions
Format
Multiple choice questions
Difficulty
Graduate level
BenchLM stores GPQA-D separately from the standardized GPQA row when providers publish exact chart values that should not overwrite the core weighted benchmark.
Freshness and provenance
Version
GPQA-D 2023
Refresh cadence
Static
Staleness state
Stale
Question availability
Public benchmark set
StaleDisplay only
BenchLM uses freshness metadata to decide whether a benchmark should still be treated as a strong differentiator, a benchmark to watch, or a display-only reference. For the full scoring policy, see the BenchLM methodology page.
Questions
What does GPQA-D measure?
A display-only GPQA Diamond reference from provider comparison charts.
Which model scores highest on GPQA-D?
GPT-6 Astra by OpenAI currently leads with a score of 96.0% on GPQA-D.
How many models are evaluated on GPQA-D?
86 AI models have published results on GPQA-D in the BenchLM catalog.
Last updated: October 29, 2026 · BenchLM version GPQA-D 2023
Know when it’s worth switching models
The model to choose, the cheaper alternative, and the release we would wait on.
Read a sample issueJoin 6,100+ subscribers.
One email each week. Unsubscribe anytime.