[GEMINI 2.5/3.1] Suitable for everyday use, but unstable with prolonged solo use.
[Claude Sonnet] Frequent issues with style distortion and errors during testing.
[Claude Opus] The model closest to the intended style and atmosphere.
[GPT] Can also output in a style close to the intended one.
[Huey/Deepseek] High token count characters, not recommended for these models.