Benchmarks · safety
UGI Willingness
A private-question evaluation of how far a model follows challenging instructions before refusing or deviating. Maintained by DontPlanToEnd. Score: W/10 👍.
167 models · 168 results · data as of 2026-10-29 · huggingface.co ↗
Leaderboard
Qwen 2.5 14B
All results
168 rows| # | Model | Effort | UGI Willingness |
|---|---|---|---|
| 1 | Qwen 2.5 14B | 8.2 | |
| 2 | Qwen 2.5 32B | none | 8.2 |
| 3 | Qwen 3 | high | 8.0 |
| 4 | Gemma 1 2B | 7.8 | |
| 5 | Gemma 2 9B | 7.8 | |
| 6 | Mistral Large 2411 | 7.5 | |
| 7 | Qwen 2.5 72B | none | 7.5 |
| 8 | Qwen 3 4B | none | 7.5 |
| 9 | Qwen 3.1 | none | 7.5 |
| 10 | DeepSeek V4 Flash | max | 7.2 |
| 11 | Mistral Medium 3 | none | 7.2 |
| 12 | Gemma 2 7B | 7.0 | |
| 13 | Gemma 3 12B | none | 7.0 |
| 14 | Llama 7 0b | 7.0 | |
| 15 | Devstral Small 2 | none | 6.8 |
| 16 | Magistral Small 2509 | 6.8 | |
| 17 | Grok 3 | none | 6.5 |
| 18 | Grok 4.1 Fast | high | 6.5 |
| 19 | Grok 4.20 | xhigh | 6.5 |
| 20 | Mistral Large 3 | none | 6.5 |
| 21 | Mistral Small 3.1 | none | 6.5 |
| 22 | Mistral Small 3.2 | none | 6.5 |
| 23 | Mistral Small | none | 6.5 |
| 24 | Qwen 3 30B-A3B | none | 6.5 |
| 25 | DeepSeek V3 | none | 6.2 |
| 26 | Grok 4.0709 | 6.0 | |
| 27 | Mistral Medium 3.5 | none | 6.0 |
| 28 | Mistral Small 3 | none | 6.0 |
| 29 | Qwen 3 8B | none | 6.0 |
| 30 | DeepSeek V3.0324 | 5.8 |