First, an important distinction. The reply token setting is a maximum, not a target. Raising it doesn't instruct the AI to write more — it just means the AI has room to, if the story calls for it. The AI writes a reply that fits the moment, and that can legitimately be short or long.
| Tier | Max tokens per reply |
|---|---|
| Free / Get A Taste | 180 |
| True Supporter / I'm All In | 300 |
You can adjust this up to your tier's maximum in Generation Settings. If replies are being cut off mid-sentence, this is the setting to check — that's a genuine ceiling problem.
Why replies get shorter as a chat goes on. The AI takes cues from the conversation so far. If your recent messages are short, or the pattern of the chat has become clipped, it mirrors that. Early in a conversation there's more to establish, so replies naturally run longer.
What helps
- Write more substantial messages yourself. The AI matches your energy — a one-line reply invites a one-line answer. This makes more difference than any setting.
- Give the story somewhere to go. A concrete action, a question, a new element in the scene.
- Ask via Director Mode, addressed to the character rather than the AI:
/cmd Narrator should reply with a minimum of 180 tokens in this roleplay
- Try a different model. Some are naturally more verbose. See article 4.10.
- Delete the short replies so they don't become the pattern the AI copies.
Was this article helpful?
That’s Great!
Thank you for your feedback
Sorry! We couldn't be helpful
Thank you for your feedback
Feedback sent
We appreciate your effort and will try to fix the article