Judge Arena: leaderboard
Atla AI evaluation of LLM-as-judge capability via human preference. 28 models rated by ELO from 4,689 pairwise votes measuring evaluation quality, consistency, and alignment with human judgments.
Metric: ELO Score. Source: huggingface.co. Saturation forecast: Estimated already saturated. 28 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Llama 3.3 70B Instruct | 1335 | #534 |
| 2 | Claude 3 Opus | 1312 | #571 |
| 3 | GPT-4o | 1308 | #343 |
| 4 | GPT-4 Turbo | 1304 | #404 |
| 5 | Claude 3 Haiku | 1286 | #803 |
| 6 | Claude 3.5 Haiku | 1282 | #568 |
| 7 | GPT-3.5 Turbo | 1271 | #865 |
| 8 | Qwen 2.5 72B Instruct Turbo | 1269 | #527 |
| 9 | Llama 3.1 405B Instruct | 1263 | #457 |
| 10 | Llama 3.1 8B Instruct | 1233 | #1035 |
| 11 | Mistral 7B Instruct (v0.3) | 1215 | #1274 |
| 12 | Claude 3.5 Sonnet | 1211 | #350 |
| 13 | QwQ 32B-Preview | 1172 | #651 |
| 14 | Mistral Instruct v0.1 7B | 1157 | #1454 |
| 15 | Qwen 2 72B Instruct | 1123 | #576 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=judge-arena · How It Works · Data refreshed daily, snapshot 2026-10-20.