Conversation limits
This guide explains the limits that apply to agent conversations in CustomGPT.
Note: The numbers below aren't fixed per plan - they come from your agent's Context Depth, a setting you choose per agent (Compact, Balanced, Extended, Rich, Max, or Custom). A higher tier lets the agent use more of your knowledge base and remember more of the conversation, at the cost of more query credits per response. See Set context depth to see all the tiers and change your agent's Context Depth.
Conversation limits
Conversation limits define how much data, such as previous messages and knowledge base content, the agent can use when generating a response. Limits vary based on your subscription plan.
Context tokens
The context token limit controls how much of the agent’s knowledge base will be provided to the agent when generating a single response.
- Standard: 2,000 tokens (≈ 1,500 words)
- Premium: 3,000 tokens (≈ 2,250 words)
- Enterprise: On demand
Conversation history tokens
This limit defines how much of the previous conversation history is sent to the agent with each new prompt.
- Standard & Premium: 1,000 tokens (≈ 750 words)
- Enterprise: On demand
User prompt length
This sets the maximum length of a user’s message.
- Standard & Premium: 8,000 characters
- Enterprise: On demand
Related articles
Updated 4 days ago