Reasoning
Multi-step logic and hard questions.
Benchmarks
The catalog, scored by category.
Sample bars for layout preview. These figures are not measured results.
| # | AI model | Composite | Score | |
|---|---|---|---|---|
| 1 | Kimi-K3-Uncensored-GGUF | 86.4 | 86.4 | |
| 2 | DeepSeek-V4.1-Flash-UNCENSORED-FP8 | 82.7 | 82.7 | |
| 3 | GLM-5.3-Flash-Uncensored-FP8 | 79.8 | 79.8 | |
| 4 | MiMo-V2.6-Pro-RL-UNCENSORED | 76.5 | 76.5 | |
| 5 | Qwen3.8-27B-Uncensored-FP8 | 73.9 | 73.9 | |
| 6 | gemma-4-31B-it-uncensored | 71.2 | 71.2 | |
| 7 | Qwen-Image-2.1-Uncensored-GGUF | 67.4 | 67.4 | |
| 8 | LTX-2.5-uncensored-v1.1-FP8 | 63.8 | 63.8 |
Overall
Same sample scores, shown together. A stronger green is a higher score.