Benchmark
Arena ELO - German
People compare model answers and choose which they prefer. Higher scores are better.
All human preference benchmarksScores from LMArena data.
- Models measured
- 273
- Scores recorded
- 319
- Sources
- 1
- IQ weight
- 1.52
Score distribution
- 1,328.6
- Mean
- 1,372.1
- Median
- 138.3
- Std dev
- 1,110.8
- 10th percentile
- 1,469.8
- 90th percentile
- 901.9
- Min
This benchmark has a weight of 1.52 for chat models, measured over 273 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 |
|---|---|---|
| 301 | Llama 3.2 3B· LMArena data | 1,056.1 |
| 302 | GPT-3.5 Turbo1106· LMArena data | 1,055.9 |
| 303 | Gemma 1.1 7B IT· LMArena data | 1,053.6 |
| 304 | Phi-3 Mini 4K Instruct· LMArena data | 1,043.9 |
| 305 | Qwen1.5-14B-Chat· LMArena data | 1,042.5 |
| 306 | Yi 34B Chat· LMArena data | 1,042 |
| 307 | Llama 2 70B Chat· LMArena data | 1,041.2 |
| 308 | Phi-3 Mini 4K InstructJune 2024· LMArena data | 1,030.8 |
| 309 | Starling LM 7B Alpha· LMArena data | 1,025.2 |
| 310 | Llama 3.2 1B Instruct· LMArena data | 1,013.6 |
| 311 | Llama 2 13B Chat· LMArena data | 1,009 |
| 312 | Phi-3 Mini 128K Instruct· LMArena data | 1,006 |
| 313 | Vicuna 13B· LMArena data | 992.3 |
| 314 | Zephyr 7B Beta· LMArena data | 989 |
| 315 | Mistral 7B Instruct v0.2· LMArena data | 987.3 |
| 316 | Llama 2 7B Chat· LMArena data | 977.5 |
| 317 | Gemma 1.1 2B IT· LMArena data | 943.8 |
| 318 | Mistral 7B Instruct· LMArena data | 931 |
| 319 | Qwen1.5-4B-Chat· LMArena data | 901.9 |
Showing 301–319 of 319 results.
Data credits: LMArena, CC BY 4.0 (licence)