Fix an AI SDK 404 on /responses.

A custom base URL does not select the API format. In the AI SDK OpenAI provider, the default model factory uses the Responses API even when you set a custom baseURL. If the server only implements Chat Completions, the SDK can send POST /v1/responses and receive a 404.

Check the request path

Capture the outgoing method and URL before changing credentials. Compare the final path with the provider's current endpoint documentation:

A 404 can also mean the host, deployment path or API version is wrong. Use the actual captured URL and the provider's documented path to distinguish those cases.

Select Chat Completions when the endpoint supports it

For a Chat Completions-only endpoint, use the provider's .chat() factory. Keep the documented base URL as a prefix and use the exact model ID available to your key.

import { createOpenAI } from "@ai-sdk/openai";
import { generateText } from "ai";

const provider = createOpenAI({
  baseURL: process.env.MODEL_API_BASE_URL,
  apiKey: process.env.MODEL_API_KEY,
});

// Use this when the endpoint documents Chat Completions.
const model = provider.chat(process.env.MODEL_ID ?? "MODEL_ID_FROM_PROVIDER");
const result = await generateText({ model, prompt: "Reply in one sentence." });
console.log(result.text);

Keep credentials in a local environment file or server-side secret store. Do not paste an API key into a browser page, public source file or support report.

Use Responses only when the endpoint supports it

Responses and Chat Completions are separate API surfaces. OpenAI's migration guide shows different endpoint paths and request/response shapes: Responses uses input and a typed output array, while Chat Completions uses messages and choices. If the host documents Responses, keep the Responses model method and confirm its supported features; otherwise select Chat Completions or the SDK's OpenAI-compatible provider.

ChinaGenAI endpoint

ChinaGenAI's site API exposes POST /v1/chat/completions. It does not expose POST /v1/responses. Use a Chat Completions client method with the site base URL, site key and free model label. Review the route and feature limits before connecting an app.

Check the source for your SDK version

• AI SDK OpenAI provider: custom base URLs and model methods
• AI SDK OpenAI-compatible providers
• OpenAI: migrate between Responses and Chat Completions
• A reported custom-baseURL 404 example (one report, not evidence of how often the issue occurs)

Documentation checked 22 October 2026. This guide verifies the request method and path from documentation; no provider account, API key or model request was used.

Free API access and limits

Check the site free API access and limits before trying to obtain a site key. Read the current status and request limits before using a site key. It is not a source of official provider keys.

ChinaGenAI is not affiliated with the providers named here. This release contains setup guidance and explicitly labeled fictional examples, not claimed live model benchmarks.