You get 550 Premium LLM Calls per user per 24 hours. The limit resets 24 hours after your first agentic request in a given period. Each paid seat gets its own call bucket that isn't shared or pooled with other users.

When you reach your daily limit, courtesy mode kicks in - responses slow down and may fall back to smaller models, but service doesn't cut off. You'll still have basic access through a throttled version of the Agent.

Your live LLM counter is displayed on your profile page. After every agent usage, you can see at the bottom of the message how many premium LLM calls were consumed, how many remain for today, and when your limit will reset.

A Premium LLM Call occurs each time the Agent interacts with the Large Language Model while resolving your task. For example, if an Agent uses 5 different tools before giving you a final response, it consumes 6 Premium Calls total (5 for tools + 1 for the final response).

Yes, you can bring your own OpenAI, Anthropic, or Gemini keys. You'll pay the Core seat fee to Zencoder, and when using BYOK, all LLM calls with that model will use your key instead of counting against your daily limits.

No, basic autocomplete is currently unlimited across all plans. Only agent-driven tasks like chat, refactor, and multi-file edits count toward your daily Premium LLM Call total.

Use separate chats for different topics to avoid confusion from long chat history. Use Basic Chat Agent for simple queries that don't require extensive code generation. Create custom agents for repetitive tasks to reduce tool calls. Avoid non-task messages like "thank you" as all messages count against limits.