Inferock Serve is managed, serverless inference with accountability.
This guide covers setup and a first inference call in the development app.
1. Activate Inferock Serve
Open the development app and sign in or create an account. Select Inferock Serve during setup. Read the Terms and Privacy Policy, accept them, then select Continue with Inferock Serve.
You do not need a provider key. Use your Inferock API key for gateway requests; Inferock manages access to the managed inference partner.
2. Save your Inferock API key
Successful activation displays your first Inferock key with a Shown once label. Save it now, before selecting Continue. This key is not shown again. Use it to authenticate your gateway requests.
If you lose the key, open Settings → Inferock keys, enter a label and select Issue key. Save the new key when it is shown, then revoke the lost key.
3. Save the gateway base URL
Activation shows the gateway base URL with your key, under First proxied call. Copy that base URL and save it for future requests. It is the value for GATEWAY_BASE_URL; the dashboard address is not the gateway base URL.
You can copy the gateway base URL at any time from Settings → Inferock keys. Select Copy beside Gateway base URL. If that row is missing, follow Support and contact.
4. Make one managed inference call
Replace the two values below with the key and gateway base URL from activation:
export INFEROCK_API_KEY='paste-your-shown-once-inferock-key'
export GATEWAY_BASE_URL='paste-the-gateway-base-url-shown-at-activation'This example uses the model from the activation command. Choose a unique request ID; the example uses inferock-serve-first-call-001. Change it before each new call so you can find that request later.
curl -sS -i "${GATEWAY_BASE_URL%/}/v1/chat/completions" \
-H "Authorization: Bearer ${INFEROCK_API_KEY}" \
-H 'Content-Type: application/json' \
-H 'x-governance-provider: inferock-serve' \
-H 'x-request-id: inferock-serve-first-call-001' \
-d '{"model":"mistralai/Mistral-Small-3.2-24B-Instruct-2506","messages":[{"role":"user","content":"Reply with one short sentence."}]}'A successful request returns HTTP 200 and the model's response.
5. Find the call
Open Calls in the dashboard and find the request ID you sent in the x-request-id header. For the example above, match inferock-serve-first-call-001. Open the call details to confirm the request ID and inspect the outcome and measured usage.
When a request fails
A managed inference error can look like this:
404 model_not_available
The requested model is not available.This means the requested model is not available. Check the model value against the activation example. If that example also fails, include the status, error text and request ID when you contact support.
Get help
Follow Support and contact for help with activation, a missing key, a missing Gateway base URL row in Settings → Inferock keys, or a failed call. Do not include your API key in a support message.