Benchmark
GPQA
Science benchmarks measure how well models answer science questions. Scores are percentages. Higher is better.
- Models and systems measured
- 364
- Scores recorded
- 378
- Sources
- 3
- IQ weight
- 0.75
Score distribution
- 45.3%
- Mean
- 47.5%
- Median
- 34.4%
- Std dev
- 2.9%
- 10th percentile
- 89%
- 90th percentile
- 0%
- Min
This benchmark has a weight of 0.75 for chat models, measured over 359 models. The weight is how much it counts toward llminfo IQ. We measure it from the data, not from the raw scores.
Leaderboard
| # | Model or system | Score |
|---|---|---|
| 101 | Claude Opus 4· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 79.6% |
| 102 | Nova 2 Lite· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 79.6% |
| 103 | Nemotron 3 Super 120B A12BNeMo Evaluator / mostly NeMo Skills · no tools· Reported by NVIDIA· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 79.2% |
| 104 | GLM 4.5· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 79.1% |
| 105 | O1 Pro· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 79% |
| 106 | Seed 2.0 Mini· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 79% |
| 107 | Gemma 4 12B It· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 78.8% |
| 108 | Sarvam 105B· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 78.7% |
| 109 | MiniMax M2· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. | 78% |
| 110 | o1· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 78% |
| 111 | Qwen3 235B A22B Instruct 2507· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 77.5% |
| 112 | o3 Mini· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 77.2% |
| 113 | Qwen3 Next 80B A3Bthinking· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 77.2% |
| 114 | Qwen3.5 Omni FlashQwen3.5-Omni release · no-thinking setting· Reported by Qwen, Qwen Team· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 76.4% |
| 115 | Qwen3.5 4B· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 76.2% |
| 116 | Llama 3.1 Nemotron Ultra 253B v1· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 76% |
| 117 | Kimi K2 0905· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 75.8% |
| 118 | Nemotron 3.5 Lightning 30B A3B· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 75.4% |
| 119 | Claude Sonnet 4· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 75.4% |
| 120 | GLM 4.7 Flash· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 75.2% |
| 121 | Kimi K2 Instruct· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 75.1% |
| 122 | GLM 4.5 Air· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 75% |
| 123 | Nemotron 3 Nano 30B A3B· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 75% |
| 124 | DeepSeek V3.1· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 74.9% |
| 125 | qwen3-vl-30bthinking· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. | 74.4% |
| 126 | Gemini 2.0 Flashthinking· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 74.2% |
| 127 | GPT OSS 20bhigh· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 74.2% |
| 128 | ERNIE 4.5· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 74% |
| 129 | Mercury 2· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 74% |
| 130 | DeepSeek R1 Zero· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 73.3% |
| 131 | O1 Preview· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 73.3% |
| 132 | LongCat Flash Chat· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 73.2% |
| 133 | Qwen3 VL 32B Thinking· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. | 73.1% |
| 134 | Claude Haiku 4.5· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 73% |
| 135 | Qwen3 Next 80B A3B Instruct· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 72.9% |
| 136 | GPT OSS 20b· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 71.5% |
| 137 | GPT-5 nano· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 71.2% |
| 138 | Ministral 14Blatest· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 71.2% |
| 139 | Mistral Smalllatest· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 71.2% |
| 140 | Magistral Medium· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 70.8% |
| 141 | qwen3-vl-30b· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. | 70.4% |
| 142 | GPT 4o· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 70.1% |
| 143 | MiniMax M1 80K· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. | 70% |
| 144 | Qwen: Qwen3 VL 8B Thinking· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. | 69.9% |
| 145 | Llama 4 Maverick· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 69.8% |
| 146 | GPT-4.5 Preview· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 69.5% |
| 147 | MiniMax M1 40K· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. | 69.2% |
| 148 | Phi 4 Reasoning Plus· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 68.9% |
| 149 | Qwen3 VL 32B Instruct· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. | 68.9% |
| 150 | Magistral Small 25092506· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 68.2% |
Showing 101–150 of 374 results.
Not ranked
The source marks these results as descriptive, not capability scores, so we list them by name instead of ranking them.
| Model or system | Score |
|---|---|
| Qwen3.5-397B-A17B-BaseQwen3.5 release/model card · base-model table· Reported by Qwen· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 54.6% |
| Qwen3.7-Plus-BaseQwen3.8-Flash-Next release/model card · base-model table· Reported by Qwen· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 51.5% |
| Qwen3.8-27B-BaseQwen3.8-Flash-Next release/model card · base-model table· Reported by Qwen· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 45% |
| Qwen3.8-Flash-Next-BaseQwen3.8-Flash-Next release/model card · base-model table· Reported by Qwen· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 51.4% |
Data credits: Data by LLM Stats · BenchBoard by SYuan03 (github.com/SYuan03/benchboard). MIT covers its code and data organization; no results licence is inferred. Numeric relays retain the original stating-source links.