Benchmark
Arena ELO Text Factuality - Industry Life and Physical and Social Science
People compare model answers and choose which they prefer. Higher scores are better.
All human preference benchmarksScores from LMArena data.
- Models measured
- 138
- Scores recorded
- 156
- Sources
- 1
- IQ weight
- 2.35
Score distribution
- 1,466.8
- Mean
- 1,468.6
- Median
- 33.7
- Std dev
- 1,433.2
- 10th percentile
- 1,506.4
- 90th percentile
- 1,348.5
- Min
This benchmark has a weight of 2.35 for chat models, measured over 138 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 | Trinity Large Thinking· LMArena data | 1,396 |
| 152 | Trinity Large Preview· LMArena data | 1,386.5 |
| 153 | Nemotron 3 Nano 30B A3BBF16· LMArena data | 1,375.8 |
| 154 | Nemotron 3.5 Lightning 30B A3BNVFP4· LMArena data | 1,374.4 |
| 155 | Command A03-2025· LMArena data | 1,367.9 |
| 156 | Olmo 3.1 32B Instruct· LMArena data | 1,348.5 |
Showing 151–156 of 156 results.
Data credits: LMArena, CC BY 4.0 (licence)