URIAL-Bench - Roleplay — leaderboard
Metric: Judge Score (0-10). Source: huggingface.co. 19 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | GPT-4 | 8.9 |
| 2 | GPT-3.5 Turbo | 8.4 |
| 3 | DBRX | 7.93 |
| 4 | Yi 34B (Base) | 7.75 |
| 5 | Mistral-7B-v0.1 | 7.65 |
| 6 | Llama 2 7B Base | 7.48 |
| 7 | Mixtral 8x7B (v0.1) | 7.4 |
| 8 | Llama 2 70B Base | 7.33 |
| 9 | Phi-2 | 7.2 |
| 10 | Yi 6B (Base) | 6.95 |
| 11 | Llama 2 13B Base | 6.63 |
| 12 | gemma-7B | 6.25 |
| 13 | gemma-2B | 5.7 |
| 14 | falcon-7B | 4.55 |
| 15 | mpt-7B | 1.9 |
Interactive version: theaggregate.ai/benchmark?slug=urial-bench-roleplay · How the rankings work · Data refreshed daily, snapshot 2026-07-22.