GPQA Diamond Leaderboard
reasoning169 models · Updated October 2026
On the GPQA Diamond benchmark, Gemini 4 Argon ranks #1 with a score of 96.4, while Qwen3 4B offers the best score-per-dollar at $0.03/1M output tokens. The full ranking, with cost per million tokens, is below.
💰 Best value
Qwen3 4B — score 52.3 at $0.03/1M output tokens
| # | Model | Score | Output / 1M | Context |
|---|---|---|---|---|
| 1 | Gemini 4 ArgonGoogle DeepMind | 96.4 | — | — |
| 2 | OpenAI: GPT-6 AstraOpenAI | 95.8 | $50.00 | 1.1M |
| 3 | Anthropic: Claude Sonnet 5.5Anthropic | 95.6 | $10.00 | 1M |
| 4 | OpenAI: GPT-6.1 SolOpenAI | 95.4 | $10.00 | 1.1M |
| 5 | Gemini 3.8 FlashGoogle DeepMind | 95.4 | — | — |
| 6 | Google: Gemini 3.7 FlashGoogle | 94.8 | $3.75 | 1M |
| 7 | OpenAI: GPT-5.4 ProOpenAI | 94.6 | $180.00 | 1.1M |
| 8 | Google: Gemini 3.1 Pro PreviewGoogle | 94.4 | $12.00 | 1M |
| 9 | OpenAI: GPT-6 SolOpenAI | 94.3 | $10.00 | 1.1M |
| 10 | Google: Gemini 3.6 FlashGoogle | 94.1 | $3.75 | 1M |
| 11 | SpaceXAI: Grok 4.6SpaceXAI | 94.0 | $6.00 | 500K |
| 12 | Anthropic: Claude Opus 5Anthropic | 93.9 | $25.00 | 1M |
| 13 | OpenAI: GPT-5.6 SolOpenAI | 93.5 | $10.00 | 1.1M |
| 14 | SpaceXAI: Grok 4.5SpaceXAI | 93.4 | $6.00 | 500K |
| 15 | OpenAI: GPT-5.6 TerraOpenAI | 93.3 | $12.00 | 1.1M |
| 16 | OpenAI: GPT-5.4OpenAI | 93.3 | $15.00 | 1.1M |
| 17 | MoonshotAI: Kimi K3MoonshotAI | 93.1 | $15.00 | 1M |
| 18 | Gemini 3.5 FlashGoogle DeepMind | 92.8 | — | — |
| 19 | SpaceXAI: Grok 4.7SpaceXAI | 92.7 | $6.00 | 500K |
| 20 | Qwen3.8 MaxAlibaba | 92.7 | $6.00 | 1M |
| 21 | Gemini 3 Pro PreviewGoogle | 92.6 | $12.00 | 1M |
| 22 | Z.ai: GLM 5.2Z.ai | 91.9 | $7.00 | 1M |
| 23 | DeepSeek: DeepSeek V4 Pro 0423DeepSeek | 91.7 | $1.91 | 1M |
| 24 | OpenAI: GPT-5.6 LunaOpenAI | 91.6 | $1.20 | 1.1M |
| 25 | OpenAI: GPT-5.2OpenAI | 91.4 | $14.00 | 400K |
| 26 | Anthropic: Claude Opus 4.8Anthropic | 91.0 | $25.00 | 1M |
| 27 | DeepSeek: DeepSeek V4 Flash 0423DeepSeek | 91.0 | $0.18 | 1M |
| 28 | GLM-5.3Z.ai (Zhipu AI) | 90.9 | — | — |
| 29 | MiniMax: MiniMax M3MiniMax | 90.9 | $1.20 | 1M |
| 30 | Qwen3.7 MaxAlibaba | 90.9 | $7.50 | 1M |
| 31 | MoonshotAI: Kimi K2.6MoonshotAI | 90.8 | $3.41 | 262K |
| 32 | OpenAI: GPT-5.5OpenAI | 90.7 | $30.00 | 1.1M |
| 33 | Anthropic: Claude Opus 5.5Anthropic | 90.6 | $20.00 | 1M |
| 34 | Anthropic: Claude Opus 4.6Anthropic | 90.5 | $25.00 | 1M |
| 35 | Anthropic: Claude Sonnet 5Anthropic | 90.5 | $10.00 | 1M |
| 36 | OpenAI: GPT-6 LunaOpenAI | 90.5 | $0.50 | 1.1M |
| 37 | GLM-5.3-FlashZ.ai (Zhipu AI) | 90.2 | — | — |
| 38 | Anthropic: Claude Opus 4.7Anthropic | 90.2 | $25.00 | 1M |
| 39 | GLM-5.1Z.ai (Zhipu AI) | 89.9 | — | — |
| 40 | DeepSeek: DeepSeek V4.1 FlashDeepSeek | 89.8 | $0.89 | 1M |
| 41 | Muse SparkMeta AI | 89.8 | — | — |
| 42 | Anthropic: Claude Haiku 5.5Anthropic | 89.7 | $0.50 | 1M |
| 43 | Google: Gemini 3 Flash PreviewGoogle | 89.4 | $3.00 | 1M |
| 44 | SpaceXAI: Grok 4.20SpaceXAI | 89.3 | $2.50 | 2M |
| 45 | SpaceXAI: Grok 4.3SpaceXAI | 88.8 | $2.50 | 1M |
| 46 | Thinking Machines: Inkling SmallThinking Machines | 88.5 | $1.20 | 524K |
| 47 | Qwen3.6 PlusAlibaba | 88.4 | $3.00 | 1M |
| 48 | Thinking Machines: InklingThinking Machines | 88.3 | $4.05 | 524K |
| 49 | Mistral: Mistral Large 4Mistral | 88.1 | $2.09 | 1M |
| 50 | MoonshotAI: Kimi K2.7 CodeMoonshotAI | 87.9 | $3.35 | 262K |
| 51 | Qwen3.7 PlusAlibaba | 87.9 | $1.60 | 1M |
| 52 | GLM-5Z.ai (Zhipu AI) | 87.8 | — | — |
| 53 | OpenAI: GPT-5.1OpenAI | 87.6 | $10.00 | 400K |
| 54 | Qwen3.6 Max PreviewAlibaba | 87.4 | $7.80 | 262K |
| 55 | Anthropic: Claude Sonnet 4.6Anthropic | 87.4 | $15.00 | 1M |
| 56 | Grok 4xAI | 87.0 | — | — |
| 57 | OpenAI: GPT-5.4 MiniOpenAI | 86.9 | $4.50 | 400K |
| 58 | Qwen3.5 397B-A17BAlibaba | 86.4 | $3.60 | 262K |
| 59 | OpenAI: GPT-5OpenAI | 86.2 | $10.00 | 400K |
| 60 | Anthropic: Claude Opus 4.5Anthropic | 86.0 | $25.00 | 200K |
| 61 | Qwen3.6 27BAlibaba | 85.9 | $3.60 | 262K |
| 62 | Anthropic: Claude Fable 5Anthropic | 85.9 | $50.00 | 1M |
| 63 | NVIDIA: Nemotron 3 UltraNVIDIA | 85.3 | $2.40 | 262K |
| 64 | Gemini 2.5 Pro (Jun 2025)Google DeepMind | 85.3 | — | — |
| 65 | Qwen3.5 PlusAlibaba | 84.8 | $2.40 | 1M |
| 66 | Qwen3.6 35B-A3BAlibaba | 84.8 | $1.49 | 262K |
| 67 | Google: Gemini 2.5 Pro Preview 06-05Google | 84.8 | $10.00 | 1M |
| 68 | Kimi K2 Thinking TurboMoonshot AI | 84.2 | $8.00 | 262K |
| 69 | Qwen3.5 35B-A3BAlibaba | 83.5 | $2.00 | 262K |
| 70 | DeepSeek ReasonerDeepSeek | 83.4 | $0.28 | 1M |
| 71 | Qwen3.6 FlashAlibaba | 83.3 | $1.13 | 1M |
| 72 | Google: Gemini 3.5 Flash LiteGoogle | 83.3 | $2.50 | 1M |
| 73 | GLM-4.7Z.ai (Zhipu AI) | 83.3 | — | — |
| 74 | GPT-5.5 InstantOpenAI | 82.5 | — | — |
| 75 | Qwen3.7 FlashAlibaba | 82.3 | $0.13 | 1M |
| 76 | Qwen3.5 FlashAlibaba | 82.3 | $0.40 | 1M |
| 77 | Anthropic: Claude Sonnet 4.5Anthropic | 82.3 | $15.00 | 1M |
| 78 | OpenAI: o3OpenAI | 81.8 | $8.00 | 200K |
| 79 | Google: Gemini 3.1 Flash LiteGoogle | 81.8 | $1.50 | 1M |
| 80 | Qwen3 235B-A22BAlibaba | 80.0 | $2.80 | 131K |
| 81 | OpenAI: o4 MiniOpenAI | 79.6 | $4.40 | 200K |
| 82 | Qwen: Qwen3.5-9BQwen | 79.0 | $0.15 | 262K |
| 83 | Claude 3.7 SonnetAnthropic | 78.5 | — | — |
| 84 | OpenAI: GPT-5.4 NanoOpenAI | 78.5 | $1.25 | 400K |
| 85 | Anthropic: Claude Sonnet 4Anthropic | 78.3 | $15.00 | 200K |
| 86 | Anthropic: Claude Opus 4.1Anthropic | 77.3 | $75.00 | 200K |
| 87 | OpenAI: o3 MiniOpenAI | 77.0 | $4.40 | 200K |
| 88 | OpenAI: o1OpenAI | 76.8 | $60.00 | 200K |
| 89 | DeepSeek-R1DeepSeek | 76.3 | — | — |
| 90 | Anthropic: Claude Opus 4Anthropic | 76.3 | $75.00 | 200K |
| 91 | Gemma 4 31B ITQVAC | 75.8 | $0.00 | 262K |
| 92 | OpenAI: gpt-oss-120b (batch)OpenAI | 75.8 | $0.14 | 131K |
| 93 | OpenAI: GPT-5 MiniOpenAI | 75.0 | $2.00 | 400K |
| 94 | Gemma 4 26B A4B ITVertex | 73.2 | $0.60 | 262K |
| 95 | Qwen3 MaxAlibaba | 72.6 | $6.00 | 262K |
| 96 | ByteDance-Seed/Seed-OSS-36B-InstructNvidia | 71.5 | $0.00 | 262K |
| 97 | Deepseek-Chat302.AI | 71.2 | $0.43 | 128K |
| 98 | Anthropic: Claude Haiku 4.5Anthropic | 71.2 | $5.00 | 200K |
| 99 | Qwen: Qwen3 30B A3BQwen | 70.1 | $0.50 | 41K |
| 100 | OpenAI: GPT-5 NanoOpenAI | 69.4 | $0.40 | 400K |
| 101 | DeepSeek: DeepSeek V3DeepSeek | 67.6 | $0.89 | 164K |
| 102 | Llama-4-Maverick-17B-128E-Instruct-FP8Vercel AI Gateway | 67.0 | $0.00 | 128K |
| 103 | OpenAI: GPT-4.1OpenAI | 66.9 | $8.00 | 1M |
| 104 | OpenAI: GPT-4.1 MiniOpenAI | 65.8 | $1.60 | 1M |
| 105 | Qwen3 32BAlibaba | 65.7 | $2.80 | 131K |
| 106 | QwQ PlusAlibaba | 65.4 | $2.40 | 131K |
| 107 | QwQ 32BAlibaba (China) | 65.3 | $0.86 | 131K |
| 108 | DeepSeek R1 Distill Qwen 32BAlibaba (China) | 64.1 | $0.86 | 33K |
| 109 | Qwen3 14BAlibaba | 63.8 | $1.40 | 131K |
| 110 | o1-miniOpenAI | 62.4 | — | — |
| 111 | OpenAI: gpt-oss-20b (batch)OpenAI | 60.8 | $0.11 | 131K |
| 112 | Z.ai: GLM 4.7 FlashZ.ai | 60.5 | $0.40 | 200K |
| 113 | Mistral Medium (latest)Mistral | 59.5 | $7.50 | 262K |
| 114 | Qwen3 8BAlibaba | 56.8 | $0.70 | 131K |
| 115 | Qwen MaxAlibaba | 56.1 | $6.40 | 33K |
| 116 | Magistral SmallMistral | 56.1 | $1.50 | 128K |
| 117 | Microsoft: Phi 4Microsoft | 56.1 | $0.14 | 16K |
| 118 | DeepSeek R1 Distill Llama 70BAlibaba (China) | 55.7 | $0.86 | 33K |
| 119 | Qwen: Qwen3 30B A3B Instruct 2507Qwen | 55.6 | $0.30 | 262K |
| 120 | Claude 3.5 SonnetAnthropic | 55.3 | — | — |
| 121 | Grok-2xAI | 53.8 | — | — |
| 122 | Qwen3 4BNovitaAI | 52.3 | $0.03 | 128K |
| 123 | Llama 4 Scout 17B 16E InstructAzure | 51.8 | $0.78 | 128K |
| 124 | Mistral LargeMistral AI | 51.3 | $6.00 | 128K |
| 125 | Llama 3.1 405B InstructCortecs | 50.9 | $0.00 | 128K |
| 126 | o1-previewOpenAI | 50.3 | — | — |
| 127 | OpenAI: GPT-4oOpenAI | 49.2 | $10.00 | 128K |
| 128 | Qwen2.5 72B InstructAlibaba | 49.1 | $5.60 | 131K |
| 129 | Mistral Small 3.2Mistral | 49.0 | $0.30 | 128K |
| 130 | OpenAI: GPT-4.1 NanoOpenAI | 48.9 | $0.40 | 1M |
| 131 | Qwen PlusAlibaba | 48.1 | $1.20 | 1M |
| 132 | Gemma 3 27B ITCortecs | 47.7 | $0.30 | 131K |
| 133 | Mistral Small (latest)Mistral | 47.5 | $0.60 | 262K |
| 134 | Meta: Llama 3.3 70B InstructMeta | 47.4 | $0.79 | 131K |
| 135 | Mistral: Mistral Small 3Mistral | 47.3 | $0.08 | 33K |
| 136 | Claude 3 OpusAnthropic | 47.2 | — | — |
| 137 | OpenAI: GPT-4 TurboOpenAI | 46.6 | $30.00 | 128K |
| 138 | Qwen2.5 32B InstructAlibaba | 46.1 | $2.80 | 131K |
| 139 | Qwen3 4B InstructPioneer | 45.8 | $0.20 | 33K |
| 140 | DeepSeek R1 Distill Qwen 14BAlibaba (China) | 44.7 | $0.43 | 33K |
| 141 | Meta: Llama 3.1 70B InstructMeta | 44.2 | $0.40 | 131K |
| 142 | WizardLM-2 8x22B | 43.4 | $0.62 | 66K |
| 143 | Mistral Small 3.1Azure | 41.9 | $0.30 | 128K |
| 144 | Qwen TurboAlibaba | 41.8 | $0.20 | 1M |
| 145 | Llama 3.2 90B Vision InstructVercel AI Gateway | 41.0 | $0.72 | 128K |
| 146 | Claude 3 SonnetAnthropic | 40.6 | — | — |
| 147 | Meta-Llama-3-70B-InstructAzure | 40.6 | $3.54 | 8K |
| 148 | Gemma 3 12B ITKilo Gateway | 39.5 | $0.15 | 131K |
| 149 | Anthropic: Claude 3.5 HaikuHelicone | 38.1 | $4.00 | 200K |
| 150 | OpenAI: GPT-4o-miniOpenAI | 37.7 | $0.60 | 128K |
| 151 | Anthropic: Claude 3 HaikuAnthropic | 36.3 | $1.25 | 200K |
| 152 | OpenAI: GPT-4OpenAI | 35.7 | $60.00 | 8K |
| 153 | Qwen2.5 7B InstructAlibaba | 35.5 | $0.70 | 131K |
| 154 | Claude 2Anthropic | 34.7 | — | — |
| 155 | Eurus-2-7B-PRIMETsinghua University,University of Illinois Urbana-Champaign (UIUC),Shanghai AI Lab,Peking University,Shanghai Jiao Tong University,CUHK Shenzhen Research Institute | 33.9 | — | — |
| 156 | DeepSeek R1 Distill Qwen 1.5BAlibaba (China) | 33.6 | $0.00 | 33K |
| 157 | Claude 2.1Anthropic | 33.0 | — | — |
| 158 | Mixtral 8x7B Instruct v0.1Cortecs | 30.6 | $0.76 | 32K |
| 159 | Open Mistral NemoMistral | 29.9 | $0.15 | 128K |
| 160 | IBM: Granite 4.0 MicroIBM | 28.3 | $0.11 | 131K |
| 161 | OpenAI: GPT-3.5 TurboOpenAI | 28.0 | $1.50 | 16K |
| 162 | Ministral 8B (latest)Mistral | 27.1 | $0.10 | 262K |
| 163 | Meta: Llama 3.1 8B InstructMeta | 27.0 | $0.08 | 131K |
| 164 | Meta-Llama-3-8B-InstructAzure | 26.1 | $0.61 | 8K |
| 165 | Ministral 3B (latest)Mistral | 25.3 | $0.04 | 131K |
| 166 | Meta: Llama 3.2 1B InstructMeta | 23.9 | $0.20 | 60K |
| 167 | Gemma 3 4B ITKilo Gateway | 23.2 | $0.10 | 131K |
| 168 | mistral-7b-instruct-v0.3Cortecs | 15.2 | $0.11 | 127K |
| 169 | DeepSeek R1 0528 Qwen3 8BNovita AI | 9.3 | $0.09 | 128K |
What does GPQA Diamond test?
This repository contains the data presented in SuperGPQA: Scaling LLM Evaluation across 285 Graduate Disciplines. Tutorials for submitting to the official leadboard coming soon 📜 License SuperGPQA is a composite dataset that ...
Frequently asked questions
Pricing is indicative — confirm with the provider before production use. Updated October 2026.