What API usage costs
Token prices, Agent API computer and browser time, and how a call's cost is worked out.
- Am I charged for calls to retired Line Code ids?No. Line Code ids answer 410 model_retired and are never billed. Switch to a SuperCosmos id.FAQ
- Am I charged for requests that fail?Requests refused up front aren't billed. Retired ids are never billed. Partly completed work is billed for what it used.FAQ
- Am I charged for the model's thinking?Yes. Output tokens include reasoning tokens. Max and Heavy think more than Flash, so they usually write more output tokens.FAQ
- Can you show an example of what an API call costs?Worked examples: a 1,000-in / 500-out Flash chat, and the docs' Max run example of 9,696 in and 38 out (≈ $0.0176).Reference
- Does a long system prompt cost money on every request?Yes. The system prompt is part of the input on every call, so a long one adds input tokens each time.FAQ
- Does setting max_tokens lower my cost?It caps the most a single answer can cost, because you never pay for more output than the cap. Short answers cost the same either way.FAQ
- Does streaming a response cost more than a normal response?No. Streaming and non-streaming calls are billed the same way: by input and output tokens at the model's price.FAQ
- Does using tool calling on the Model API cost extra?No separate fee. Tool definitions and results are part of the input and output tokens, which are billed normally.FAQ
- How can I spend less on API calls?Use Flash by default, move up only when needed, keep context short, stream instead of polling, and set team limits.How-to
- How do I budget for thousands of API requests?Measure a typical request's tokens, multiply by the price and by your volume, then add a buffer. Check with real usage.How-to
- How do I estimate what an Agent API run will cost?Add tokens at the tier's price plus computer minutes, actions, browser minutes and storage. Try it in the Playground first.How-to
- How do I know which price list applies?The Rates page shows a rates version. Credits are charged at the prices in effect when you use them.FAQ
- How is an Agent API run on tier "auto" billed?A run with tier auto is billed at Flash rates. Runs set to max or heavy bill at those tiers.FAQ
- How is the cost of a Model API call worked out?Input tokens × input price + output tokens × output price for the model id you called, rounded up to the next millionth of a dollar.Explainer
- How much does SuperCosmos cost per token?Per 1M tokens in/out: Flash $1.60/$3.40, Max $1.80/$3.80, Heavy $2.00/$4.40. Input above 262K tokens bills 1.5×.Reference
- If I cancel an Agent API run, do I still pay?You pay for the tokens, computer time and actions used up to the cancellation. Nothing after that.FAQ
- Is SuperCosmos expensive to use through the API?Prices start at $1.60 per million input tokens. Typical short calls cost a fraction of a cent; long agent runs cost more.FAQ
- Is SuperCosmos the same price in every product?Yes. The same per-token rates apply on the Agent API, Model API, Line CLI and Cosmos Agent.FAQ
- What counts as cloud computer time on an Agent API run?Time the run's cloud computer spends running commands and file work, billed at $0.006 per minute.FAQ
- What does an Agent API run cost besides tokens?Cloud computer $0.006/min, edits and tool actions $0.0005 each, cloud browser $0.005/min, storage writes $0.10/GB.Reference
- What if I send a prompt close to the 1M-token context window?It works up to the window, but input above 262K tokens bills at 1.5×, and Model API requests are capped at 3 MB and 800 messages.FAQ
- What is a token in Clatter Labs pricing?A token is a small piece of text, roughly part of a word. Models read input tokens and write output tokens, and prices are per 1M tokens.Explainer
- Which tier am I billed at in each product?Agent API: the run's tier (auto = Flash). Line CLI: the tier that answered. Model API: the id you called. Cosmos Agent: named or inferred.Reference
- Why are my input tokens a bit higher than my prompt?Every Model API request includes a short Clatter Labs instruction block, which counts as input tokens.FAQ
- Why did the same question cost different amounts twice?Answers vary in length, chats grow longer, and different tiers have different prices. Each call bills its own usage.FAQ
- Why do long prompts cost more?A request whose prompt is over 262K tokens bills its whole input at 1.5× the normal rate.FAQ
Still need help?
Tried the steps above and still stuck? Email us. We usually reply within the first hour, and always within 24 hours. A full support team is with you.
Infrastructure questions, or permanent deletion of your data
infrastructure@clatterlabs.xyzFor deleting your whole account or the data of one service.