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 |
|---|---|---|
| 151 | Ministral 8Blatest· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 66.8% |
| 152 | longcat-flash-lite· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 66.8% |
| 153 | MiMo V2.5 Pro· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 66.7% |
| 154 | Llama 3.3 Nemotron Super 49B v1· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 66.7% |
| 155 | Sarvam-30B· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 66.5% |
| 156 | Granite 4.2 30B· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 66.4% |
| 157 | GPT 4.1· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 66.3% |
| 158 | Phi-4-reasoning· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 65.8% |
| 159 | Qwen3 30B A3B· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 65.8% |
| 160 | DeepSeek R1 Distill Llama 70b· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 65.2% |
| 161 | QwQ 32B· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claimValues from 2 sources
| 65.2% |
| 162 | QwQ 32B Preview· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claimValues from 2 sources
| 65.2% |
| 163 | GPT-4.1 mini· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 65% |
| 164 | Gemini 2.5 Flash-Lite· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 64.6% |
| 165 | Granite 4.2 8B· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 64.1% |
| 166 | Nemotron Nano 9b V2· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 64% |
| 167 | DeepSeek R1 Distill Qwen 32B· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claimValues from 2 sources
| 62.1% |
| 168 | Gemini 2.0 Flash· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 62.1% |
| 169 | Qwen3 Max· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 62% |
| 170 | O1 Mini· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 60% |
| 171 | Claude 3.5 Sonnet· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 59.4% |
| 172 | DeepSeek R1 Distill Qwen 14B· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claimValues from 2 sources
| 59.1% |
| 173 | DeepSeek V3· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 59.1% |
| 174 | Gemini 1.5 Pro· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 59.1% |
| 175 | Gemma 4 E4B-it· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 58.6% |
| 176 | Llama 4 Scout· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 57.2% |
| 177 | Phi 4· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claimValues from 2 sources
| 56.1% |
| 178 | Grok 2· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 56% |
| 179 | Granite 4.2 3B· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 54.8% |
| 180 | Ministral 3Blatest· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 53.4% |
| 181 | Aihubmix Phi 4 Mini (reasoning)· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 52% |
| 182 | Qwen3.5 2B· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 51.6% |
| 183 | Gemini 2.0 Flash Lite· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 51.5% |
| 184 | Gemini 1.5 Flash· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 51% |
| 185 | Grok 2 mini· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 51% |
| 186 | Llama 3.1 405B Instruct· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 50.7% |
| 187 | Llama 3.3 70B Instruct· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claimValues from 2 sources
| 50.5% |
| 188 | Claude 3 Opus· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 50.4% |
| 189 | GPT-4.1 nano· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 50.3% |
| 190 | Qwen2.5 32B Instruct· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claimValues from 2 sources
| 49.5% |
| 191 | DeepSeek R1 Distill Qwen 7B· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claimValues from 2 sources
| 49.1% |
| 192 | DeepSeek R1 Distill Llama 8B· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 49% |
| 193 | Qwen2.5 72B Instruct· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claimValues from 2 sources
| 49% |
| 194 | GPT-4 Turbo· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 48% |
| 195 | Qwen 3 235B A22B· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 47.5% |
| 196 | Nova Pro· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 46.9% |
| 197 | Llama 3.1 70B Instruct0-shot· Reported by Meta· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 46.7% |
| 198 | Mistral Small 3.2 24B Instruct2506· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 46.1% |
| 199 | Mistral Small 3.1 24b Instruct 2503· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 46% |
| 200 | Qwen 2.5 14B Instruct· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claimValues from 2 sources
| 45.5% |
Showing 151–200 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.