Benchmark
Arena ELO Text Factuality - Industry Writing and Literature and Language
People compare model answers and choose which they prefer. Higher scores are better.
All human preference benchmarksScores from LMArena data.
- Models measured
- 154
- Scores recorded
- 173
- Sources
- 1
- IQ weight
- 2.19
Score distribution
- 1,424.1
- Mean
- 1,427.8
- Median
- 44.3
- Std dev
- 1,368.5
- 10th percentile
- 1,472.9
- 90th percentile
- 1,271.6
- Min
This benchmark has a weight of 2.19 for chat models, measured over 154 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 |
|---|---|---|
| 151 | GPT 5.4 NanoHigh· LMArena data | 1,379.3 |
| 152 | MiMo V2 FlashThinking· LMArena data | 1,379 |
| 153 | Solar Pro 4· LMArena data | 1,374.7 |
| 154 | GLM 4.5 Air· LMArena data | 1,374.1 |
| 155 | Qwen3 Next 80B A3B Instruct· LMArena data | 1,371.5 |
| 156 | MiniMax M2.5· LMArena data | 1,368.5 |
| 157 | nova-experimental-chat10-20· LMArena data | 1,365.7 |
| 158 | GPT-5 miniHigh· LMArena data | 1,361.6 |
| 159 | Trinity Large Preview· LMArena data | 1,358.3 |
| 160 | Nemotron 3 Super 120B A12B· LMArena data | 1,350.3 |
| 161 | Trinity Large Thinking· LMArena data | 1,350.1 |
| 162 | Nova 2 Lite· LMArena data | 1,347.6 |
| 163 | Command A03-2025· LMArena data | 1,347.2 |
| 164 | GPT OSS 120B· LMArena data | 1,341.8 |
| 165 | GLM 4.7 Flash· LMArena data | 1,340 |
| 166 | Granite 4.2 30B· LMArena data | 1,338.7 |
| 167 | Nemotron 3.5 Lightning 30B A3BNVFP4· LMArena data | 1,331 |
| 168 | Granite 4.2 8B· LMArena data | 1,314.4 |
| 169 | MiniMax M2· LMArena data | 1,313.5 |
| 170 | Nemotron 3 Nano 30B A3BBF16· LMArena data | 1,311.2 |
| 171 | Olmo 3.1 32B Instruct· LMArena data | 1,297.7 |
| 172 | Olmo 3.1 32B Think· LMArena data | 1,286.6 |
| 173 | Granite 4.2 3B· LMArena data | 1,271.6 |
Showing 151–173 of 173 results.
Data credits: LMArena, CC BY 4.0 (licence)