Documentation

Quick fixes

Match your error to a solution. Most issues are resolved in under a minute with this reference table.

Error reference table

SymptomRecommended fix
401 UnauthorizedConfirm the API key is valid, includes the mr_live_ prefix, and hasn't been revoked.
403 ForbiddenThe key belongs to another project or is inactive. Issue a new key for this integration.
Provider credentials not configuredAdd the provider API key under Settings → Providers for the current project.
Structured output parsing errorsValidate your JSON schema, reduce complexity, or try a more capable model. Review the raw log entry to see the provider response.
Timeouts / slow responsesConfigure workflow fallbacks so ModelRiver can retry with alternative providers. Monitor Observability for latency patterns.
Unexpected cached dataEnsure cache fields match the payload structure. Array indexes start at 0 (for example, messages.0.content).
Webhook not receivedVerify the endpoint URL is correct and returns 2xx. Check Delivery & retries for retry status.
Webhook signature mismatchEnsure the secret matches exactly (no extra spaces). Verify you're using the raw request body for signature computation.
Rate limit exceededModelRiver enforces platform-wide rate limits. Contact support if you need higher throughput.
Empty AI responseCheck that the provider has sufficient context. Verify the messages array format and model availability.

Authentication quick fixes

API key issues

Bash
# Verify your key format
echo $MODELRIVER_API_KEY
# Should start with mr_live_
 
# Test with curl
curl -H "Authorization: Bearer $MODELRIVER_API_KEY" \
https://api.modelriver.com/v1/ai \
-d '{"workflow": "test", "messages": [{"role": "user", "content": "hello"}]}'

Provider credential issues

  1. Navigate to Settings → Providers
  2. Verify the provider shows a green checkmark
  3. If not, re-enter the credential and save
  4. Test with a simple request in the Playground

Next steps