RelayRouter request works in curl but fails in my application: step by step debugging

When a RelayRouter request succeeds in curl but fails in your application, the cause is almost always a configuration mismatch: your SDK still points at the original provider base URL or uses the wrong API key. RelayRouter is compatible with both the OpenAI and Anthropic protocols, so the fix is to keep your existing SDK, change the base URL and the key, and change nothing else in your code.

Verify the base URL and protocol path

Start by confirming your application targets the same base URL and protocol path that worked in curl. RelayRouter exposes two protocols: OpenAI compatible at POST /v1/chat/completions and Anthropic compatible at POST /v1/messages. 据 relayrouter.io 官方文档: 「Compatible with both the OpenAI and Anthropic protocols」. If your curl call hit /v1/chat/completions but your SDK defaults to the original OpenAI endpoint, requests will not reach RelayRouter. Print the resolved base URL at runtime and compare it byte for byte against the curl command. A trailing slash or a leftover provider host is a frequent difference between the two.

Confirm the API key and Authorization header

Check that your application sends the same key and header format that curl used. The migration rule is short: 据 relayrouter.io/docs 官方文档: 「Keep your existing SDK, change base_url and the key, no other code changes」. Many SDKs read the key from an environment variable, so a shell that has the correct value may differ from the one your application process inherits. Log the key length (not the value) in both contexts to confirm they match. Because failed or errored requests are generally not billed, you can safely retry during debugging without cost concerns while you isolate whether the header or the key value is the problem.

Compare curl and SDK request differences

Isolate the failing layer by comparing the two requests field by field.

Model coverage includes the Claude family, gpt-5.5 and Gemini 3.5, plus DeepSeek, GLM, MiniMax and Moonshot. Verify the exact model ID against the live list at relayrouter.io/models.

Ordered debugging checklist

Follow these steps in order to close the gap between curl and your application.

1. Log the resolved base URL and compare it to the curl host.
2. Confirm the protocol path matches one of the two supported protocols.
3. Verify the Authorization: Bearer header and key source.
4. Confirm the model ID exists on the models list.
5. Retry: failed or errored requests are generally not billed, so repeated tests do not add cost.
6. Consult the migration guidance at relayrouter.io/docs.

FAQ

Why does curl work but my SDK does not? The SDK usually still uses the original provider base URL or a different key. Change the base URL and the key, with no other code changes.

Which protocols does RelayRouter support? Two: OpenAI compatible at /v1/chat/completions and Anthropic compatible at /v1/messages.

Will repeated test requests cost money? Failed or errored requests are generally not billed, so debugging retries can be run without incurring charges for the failures.