Stick To Your Role!: leaderboard
Metric: Cardinal Score (-1 to 1). Source: huggingface.co. Saturation forecast: Estimated already saturated. 31 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Qwen 2.5 VL 72B Instruct | 0.84 |
| 2 | Llama 3.1 Nemotron 70B Instruct | 0.81 |
| 3 | Mistral Large 2 (Jul) | 0.79 |
| 4 | Llama 3.3 70B Instruct | 0.78 |
| 5 | Llama 3.1 70B Instruct | 0.77 |
| 6 | QwQ-32B | 0.77 |
| 7 | Qwen 3 32B | 0.74 |
| 8 | Mistral Large 2 (Nov) Instruct (2411) | 0.73 |
| 9 | Qwen 3 8B | 0.72 |
| 10 | Qwen 3 235B A22B FP8 | 0.72 |
| 11 | Mistral Small 3.1 | 0.7 |
| 12 | Qwen2.5-14B-Instruct-1M | 0.7 |
| 13 | Qwen 3 4B | 0.7 |
| 14 | Llama 3.1 8B Instruct | 0.62 |
| 15 | Llama 4 Scout Instruct | 0.62 |
Interactive version: theaggregate.ai/benchmark?slug=stick-to-your-role · How It Works · Data refreshed daily, snapshot 2026-09-26.