
SpicyChat AI models: access, switching and response style
A current guide to SpicyChat AI conversation models, plan access, switching models and solving repetition, language and reply-length problems.
These are conversation models
SpicyChat uses large language models to write character replies. Official documentation says more than ten models are available across different tiers. The current list changes, so the model selector inside Generation Settings is the authoritative place to see what your account can use.
Model access by plan
Free and Get A Taste use the standard selection. True Supporter unlocks additional advanced models and chat styles. I’m All In includes all models, including SpicyXL, plus generation controls such as temperature, Top-P and Top-K.
Choose by conversation behavior
| Need | What to compare | Test |
|---|---|---|
| Stronger character voice | Definition adherence | Send the same scene prompt in a fresh branch. |
| Longer prose | Verbosity and reply ceiling | Compare on the same plan; True Supporter and above allow up to 300 reply tokens. |
| Better language stability | Preferred language behavior | Set Preferred Chat Language, then test two models. |
| Break a loop | Pattern recovery | Delete repeats, switch model for 2–3 messages, then switch back. |
Do not mistake model switching for memory
A different model can change tone, length and adherence, but it does not restore messages that have already left the context window. Use Memory Manager for persistent story facts and the model selector for response behavior.