Migration from OpenAI

Overview#

Most OpenAI-compatible clients only need two changes: the base URL and the API key. Model IDs may also need to be adjusted depending on provider availability.

Step 1: change the base URL#

const client = new OpenAI({
  apiKey: process.env.UOUODUO_API_KEY,
  baseURL: "https://uouo.cloud/v1",
});

Step 2: change the API key#

Create a key in /app/keys and store it as a server-side secret:

export UOUODUO_API_KEY="sk-xxx..."

Step 3: verify model IDs#

Open /models or call /v1/models to confirm model IDs. If a direct OpenAI model ID is unavailable, choose an equivalent model from another provider.

Behavioral differences#

• Routing may choose a different upstream provider than your previous direct integration.
• Streaming chunks follow OpenAI-compatible SSE format, but provider-specific details can vary.
• Usage and cost should be verified in /app/logs.
• Tool calling and structured output support can differ by model.

Migration checklist#

1. Update base URL and API key.
2. Run a short non-streaming request.
3. Run a streaming request.
4. Test error handling for 401, 429, and 5xx.
5. Review logs, usage, and cost.
6. Roll out to staging before production.