Support OpenAI-compatible providers (Kimi, DeepSeek) with chat-completions API

[yandong2023]

Problem

OpenCode uses /v1/responses endpoint by default, but many popular providers (Kimi, DeepSeek, Gemini via OpenAI-compatible mode) only support /v1/chat/completions, resulting in 404 errors.

Affected Providers

• Kimi (Moonshot): Returns 404 for /v1/responses - the endpoint does not exist
• DeepSeek: Authentication fails due to endpoint mismatch
• Gemini (OpenAI-compatible): Similar issues with endpoint compatibility

Error Examples

Kimi

POST https://api.moonshot.cn/v1/responses → 404 Not Found

DeepSeek

Unauthorized: Authentication Fails (governor)

Proposed Solution

Add apiType configuration option to allow providers to specify which API endpoint to use.

Example Configuration

{
  "provider": {
    "kimi": {
      "npm": "@ai-sdk/openai-compatible",
      "apiType": "chat-completions",
      "options": {
        "baseURL": "https://api.moonshot.cn/v1"
      }
    }
  }
}

Backward Compatibility

Default to apiType: "responses" to maintain existing behavior for OpenAI and other providers.

Workaround

Currently users have to use OpenAI-compatible SDK with custom configuration, but the model selection and provider detection don't work properly.

Related

This affects the growing ecosystem of OpenAI-compatible APIs in China and other regions where providers have not yet migrated to the responses API.

Environment

• OpenCode version: 1.2.27
• OS: Linux
• Providers tested: Kimi (Moonshot), DeepSeek, Gemini

[github-actions]

This issue doesn't fully meet our contributing guidelines.

What needs to be fixed:

• No issue template was used. This appears to be a feature request, but it doesn't use the Feature Request template (title should start with [FEATURE]:, and the template's verification checkbox and description fields must be filled in).

Please edit this issue to address the above within 2 hours, or it will be automatically closed.

---

This issue might be a duplicate of existing issues. Please check:

• [Feature Request] Support traditional Chat Completions API for OpenAI provider #7793: Requests support for traditional Chat Completions API for the OpenAI provider — same core problem (providers only support /v1/chat/completions, not /v1/responses)
• [FEATURE]: enable responses API in openai compatible provider #9628: Requests enabling the Responses API option in openai-compatible providers — inverse of the same issue
• Azure AI Foundry/Kimi K2.5: "This model is not supported by the Responses API" #12949: Kimi K2.5 fails with "This model is not supported by the Responses API" — same provider, same root cause
• [FEATURE]: Add Provider Type in custom provider #11692: Requests a configurable Provider Type (OpenAI-compatible, Anthropic-compatible, etc.) in custom providers — overlapping solution space

If you believe this was flagged incorrectly, please let a maintainer know.

[github-actions]

This issue has been automatically closed because it was not updated to meet our contributing guidelines within the 2-hour window.

Feel free to open a new issue that follows our issue templates.

[rekram1-node]

ai-sdk/openai-compatible doesn't support responses api, so that shouldn't happen unless u are using ai-sdk/openai