Benchmark
Arena Hard
People compare model answers and choose which they prefer. Scores are percentages.
All human preference benchmarksScores from LLM Stats data.
- Models measured
- 22
- Scores recorded
- 22
- Sources
- 1
- IQ weight
- 1.10
Score distribution
- 66.4%
- Mean
- 73.3%
- Median
- 20.6%
- Std dev
- 37.9%
- 10th percentile
- 91.7%
- 90th percentile
- 30.5%
- Min
This benchmark has a weight of 1.10 for chat models, measured over 22 models. The weight is how much it counts toward llminfo IQ. We measure it from the data, not from the raw scores.
Leaderboard
| # | Model | Score |
|---|---|---|
| 1 | Qwen 3 235B A22B· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 95.6% |
| 2 | Qwen3 32B· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 93.8% |
| 3 | Qwen3 14B· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 91.7% |
| 4 | Qwen3 30B A3B· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 91% |
| 5 | 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 | 88.3% |
| 6 | Mistral Small 24B Instruct2501· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 87.6% |
| 7 | Qwen2.5 72B Instruct· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 81.2% |
| 8 | Phi 4 Reasoning Plus· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 79% |
| 9 | DeepSeek V2.5· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 76.2% |
| 10 | Phi 4· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 75.4% |
| 11 | Phi-4-reasoning· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 73.3% |
| 12 | Jamba 1.5 Large· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 65.4% |
| 13 | Mistral Smalllatest· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 58.3% |
| 14 | Ministral 3 14B Instruct2512· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 55.1% |
| 15 | Mistral Large 3· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 55.1% |
| 16 | Qwen2.5 7B Instruct· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 52% |
| 17 | Ministral 3 8b Instruct· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 50.9% |
| 18 | Jamba 1.5 Mini· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 46.1% |
| 19 | Mistral Small 3.2 24B Instruct2506· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 43.1% |
| 20 | phi-3.5-moe-instruct· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 37.9% |
| 21 | Phi 4 Mini· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 32.8% |
| 22 | Ministral 3 3b Instruct· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 30.5% |
Data credits: Data by LLM Stats