Connect your application

Create an account or sign in, then open Access keys in your workspace. Use your routr API key for requests, not your workspace session credential.

Keep your key in your server-side environment. Do not put it in browser code or commit it to source control. The examples use a placeholder: replace it locally with your own API key.

Open Credits & billing and check your available credits before making a request. Reserved credits are not available to spend. If your balance is too low, add credits first.

If you use an Unlimited plan, check that it is active and covers your chosen model. Its limits and fair-use conditions still apply. Review plans and pricing before purchasing.

Find an exact model ID in the model catalog. Check its context limit and current rates. You can also list models with your API key:

curl https://api.routr.cloud/v1/models \
  -H "Authorization: Bearer $ROUTR_KEY"

The request below uses an example model. Replace it with an ID from the current catalog; a model shown in documentation is not a promise of availability.

Set your client's base URL to:

https://api.routr.cloud/v1

Send your API key in the Authorization: Bearer header. The chat completions path is /chat/completions under that base URL.

Using an SDK? Follow the Python and Node examples. For a coding tool using Anthropic Messages, use the Anthropic setup instead; its base URL differs.

Example request: replace the API key placeholder and model ID, then run it in your own terminal. This sends a request using your balance or plan; copying the example does not execute it.

curl https://api.routr.cloud/v1/chat/completions \
  -H "Authorization: Bearer $ROUTR_KEY" \
  -H "Content-Type: application/json" \
  -d '{
  "model": "deepseek-v4-pro",
  "messages": [
    { "role": "user", "content": "Explain PKCE in two sentences." }
  ]
}'

If the request fails, read the actual error before retrying. Error codes cover invalid keys, insufficient balance, missing models, and rate limits. No example response is shown as a successful request.

Open Usage and find your request by its time and model. Inspect the outcome, token counts, and charged credits. Check the date range and filters if it is not visible.

Use the recorded outcome and actual charge to verify what happened, rather than assuming that a copied example or a catalog entry means the request succeeded.