Production and companies
Running the Model API in production and across a team.
- Can I send my customers' data through the Model API?Read the Privacy Policy and Terms first, send only what's needed, and remember you're responsible for end-user text.FAQ
- Does the Model API have an SLA?No SLA is published in the developer docs. Check the Terms of Service and contact the team for business needs.FAQ
- How can a company keep Model API costs predictable?Default to Flash, route only hard requests to Max or Heavy, cap max_tokens, trim history, and monitor usage.How-to
- How can my team share one Model API bill?Invite people to your team in Console › Team; members use the owner's credits within access and limits the owner sets.How-to
- How do I A/B test Flash against Max in production?Route a share of a request type to each model, log quality signals, usage and latency per model, and compare.How-to
- How do I avoid my app stopping when credits run out?Monitor the balance with GET /v1/account, alert at a threshold, and top up before it reaches zero.How-to
- How do I give each of my users a spending budget?Track usage per user from each response, convert it to cost with the model's prices, and stop or slow users at their limit.How-to
- How do I know if the Model API is having problems?Check status.clatterlabs.com, subscribe to email alerts or RSS, and watch your own 5xx rate.How-to
- How do I monitor abuse warnings across my app?Log x-clatter-notice on every response with your user id, alert on the first warning, and block that user quickly.How-to
- How should a SaaS product use the Model API for many customers?One server-side key per service, your own per-customer limits and logs, input moderation, and budget alerts.How-to
- How should I separate development and production?Use separate keys for each environment, with lower rate limits in development, and label them clearly.How-to
- How should my app degrade if Heavy is slow or busy?Set timeouts, stream, and fall back to Max or Flash for the same request when Heavy is overloaded or too slow.How-to
- Should I cache Model API responses?For repeated identical questions, caching on your side saves cost and time. Don't cache personal or time-sensitive answers.FAQ
- We're a company planning heavy usage. Who do we talk to?Write to support@clatterlabs.com or team@clatterlabs.xyz with your expected volume; infrastructure questions go to infrastructure@clatterlabs.xyz.FAQ
- What should I check before going to production?Keys on the server, retries with backoff, error handling, cost monitoring, abuse handling and status alerts.How-to
- What should I log for each Model API call?Log the X-Request-Id, completion id, model, usage, latency, finish_reason, error code and your own user id.Reference
- What should my app do if the model stays overloaded?Back off with Retry-After, cap retries, show a friendly message, and check the status page.How-to
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.