Best AI Model for Villain Roleplay
Sustained, calculated menace without softening or moralizing, plus honoring an out-of-character request.
Updated
Villain ranking
Score out of 100 · Higher is better
+ View as table
| Rank | Model | Score |
|---|---|---|
| 1 | DeepSeek-V4.1-Flash | 82.5 |
| 1 | MiMo-V2.6-Pro | 82.5 |
| 3 | GLM-5.3 | 80.8 |
| 4 | Kimi-K3 | 80.0 |
| 5 | GLM-5.3-Flash | 79.2 |
| 6 | GPT-6-Luna | 75.8 |
| 7 | MiMo-V2.6-Flash-MOPD | 71.7 |
| 8 | MiMo-V2.5 | 65.0 |
| 8 | GLM-4.7 | 65.0 |
| 10 | Muse-Spark-1.3 | 57.5 |
| 11 | Qwen3.8-27B | 55.8 |
| 11 | Gemma-4-31B-it | 55.8 |
| 13 | Qwen3.8-Flash-Next | 54.2 |
| 14 | DeepSeek-V4-Flash-0731 | 44.2 |
DeepSeek-V4.1-Flash and MiMo-V2.6-Pro tied for first (82.5): DeepSeek played the most controlled villain, and MiMo-V2.6-Pro's never gave away his hidden weakness. GLM-5.3 (80.8) and Kimi-K3 (80.0) followed. When the user asked out of character for shorter replies, only Kimi-K3, MiMo-V2.6-Pro, Muse-Spark-1.3, Qwen3.8-27B and GLM-4.7 were still following the request a turn later, and Muse-Spark-1.3 did so by treating it as the prisoner speaking. Gemma-4-31B-it's villain lost patience and turned to physical force, and DeepSeek-V4-Flash-0731 finished last after narrating its villain in the second person.
- 1DeepSeek-V4.1-Flash82.5
- 1MiMo-V2.6-Pro82.5
- 3GLM-5.380.8
- 4Kimi-K380.0
- 5GLM-5.3-Flash79.2
- 6GPT-6-Luna75.8
- 7MiMo-V2.6-Flash-MOPD71.7
- 8MiMo-V2.565.0
- 8GLM-4.765.0
- 10Muse-Spark-1.357.5
- 11Qwen3.8-27B55.8
- 11Gemma-4-31B-it55.8
- 13Qwen3.8-Flash-Next54.2
- 14DeepSeek-V4-Flash-073144.2
Villain: pass or fail
The specific behaviours this category tests, model by model.
Checks in this category
Models in Roleplay Index order · What happened, in a word or two
| Model | Kept an out-of-character request |
|---|---|
| 1Kimi- | Kept |
| 2GLM- | Reverted |
| 3DeepSeek- | Reverted |
| 4GLM- | Reverted |
| 5MiMo- | Kept |
| 6Muse- | Kept |
| 7GPT- | Reverted |
| 8Qwen3.8- | Reverted |
| 9MiMo- | Reverted |
| 10MiMo- | Reverted |
| 11GLM- | Kept |
| 12Qwen3.8- | Kept |
| 13Gemma- | Reverted |
| 14DeepSeek- | Reverted |
+ View as table
| Check | Kimi-K3 | GLM-5.3 | DeepSeek-V4.1-Flash | GLM-5.3-Flash | MiMo-V2.6-Pro | Muse-Spark-1.3 | GPT-6-Luna | Qwen3.8-Flash-Next | MiMo-V2.5 | MiMo-V2.6-Flash-MOPD | GLM-4.7 | Qwen3.8-27B | Gemma-4-31B-it | DeepSeek-V4-Flash-0731 |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Kept an out-of-character request | Pass: Kept | Fail: Reverted | Fail: Reverted | Fail: Reverted | Pass: Kept | Pass: Kept | Fail: Reverted | Fail: Reverted | Fail: Reverted | Fail: Reverted | Pass: Kept | Pass: Kept | Fail: Reverted | Fail: Reverted |