Output control lets you adjust reply length (output volume) and thinking (reasoning) depth directly with sliders.
All BabeChat models reply with up to 1,500 tokens per conversation at the default setting (1x, Fast). With output control you can get longer, more detailed replies, and ProChats are only deducted additionally for the portion exceeding the guaranteed base of 1,500 tokens.
If you don't use output control, no extra cost is incurred.
Conditions of Use
Available once your ProChat balance (including event ProChats) is 1,000 or more.
However, when using [fan-made bots], 'purchased ProChats' are not counted toward the balance under our policy.
Event ProChats can also be used.
Use the output control sliders in the input area to adjust output volume and thinking (reasoning) separately.
The higher the multiplier, the longer and more detailed the replies.
Setting | Description |
|---|---|
1x | Default length (up to 1,500 tokens) |
1.25x | About 1.25x length (up to 1,875 tokens) |
1.5x | About 1.5x length (up to 2,250 tokens) |
2x | About 2x length (up to 3,000 tokens) |
3x | About 3x length (up to 4,500 tokens) |
4x | About 4x length (up to 6,000 tokens) |
Raising the thinking setting makes the AI go through a deeper reasoning process before answering, producing more refined, higher-quality replies.
Setting | Description |
|---|---|
Fast | Quick responses |
Normal | Standard-level reasoning |
Deep | In-depth reasoning |
Within the default setting range, you can use it freely at no extra cost.
ProChats are deducted only for the portion exceeding the guaranteed amount.
Some models automatically perform an internal thinking (reasoning) process when replying, which also consumes tokens. Accordingly, the guaranteed tokens are split between reply output and thinking.
Category | Guaranteed Tokens |
|---|---|
output volume | 1,300 tokens |
Thinking | 200 tokens |
Total | 1,500 tokens |
Tokens exceeding the guaranteed amount are calculated in 100-token units and deducted as ProChats.
The deduction is finalized based on the tokens actually used, after the AI's response is complete.
Cost Notes
Even if you raise the output or thinking setting, the actual cost is not always higher than at a lower setting. ProChat deduction is calculated based on the tokens the AI actually uses, not the setting value — so if the AI answers concisely at a high setting, it can actually cost less than a lower setting. The multiplier widens the output ceiling; it does not mean that many tokens are consumed every time.
1. Start a 'new chat' after changing settings.
If you change settings mid-conversation, the AI may get confused with the previous conversation rules and the expected effect may not appear.
The most reliable way is to adjust the sliders before starting a new story with a character.
2. The multiplier is the AI's 'target'.
Multipliers like 1.5x and 2x are reference points the AI tries its best to match.
However, depending on the conversation's context and topic, replies may be a bit shorter or longer than the target for the most natural flow. Please be understanding — this is the AI's own judgment for the quality of the story.
3. 'Thinking depth' varies with the conversation
Like people, the AI thinks harder in conversations that require complex situations or deep emotional description.
Even at the same setting, the tokens consumed internally can vary slightly depending on how deep the topic is. This is the AI's thinking process to give you optimal intelligence.