Billing and usage

• Every workspace gets $0.50 of credit on signup.
• Runs are metered at the model's token price (Claude Haiku 5.5: $0.10 per million input tokens, $0.50 per million output; prompts over 100K tokens at $0.50 / $2.50).
• Each decided action costs $0.02, including each tool call during a run.
• Reads, results, approvals and the dashboard are free.
• Add any whole-dollar amount of credit under Billing; it doesn't expire.
• Set a monthly spend limit; requests over it return 402 spend_limit_reached.
• At zero balance, metered requests return 402 insufficient_credit and nothing is charged.

Model usage is recorded after each model call, so a run that starts with a small positive balance can end slightly below zero. The next metered request then returns 402.