API Reference

OpenAI-compatible chat completions and models endpoints. SSE streaming, hard-pinned to deepseek-v3.2.

Mikan Cloud speaks the OpenAI Chat Completions wire format. Anything you'd send to https://api.openai.com/v1 works against https://api.mikancloud.com/v1, with two pins.

`POST /v1/chat/completions`

Mirrors the OpenAI Chat Completions spec — same request body, same response, same SSE event format.

Pins

Field	Value	Notes
`model`	`deepseek-v3.2`	Any other value returns `400 model_not_found`.
`stream`	`true` recommended	Non-streaming works, but Cursor expects SSE.
`Authorization`	`Bearer sk-mk-…`	Required on every request.

Request

{
  "model": "deepseek-v3.2",
  "stream": true,
  "messages": [
    {"role": "system", "content": "You are helpful."},
    {"role": "user", "content": "ping"}
  ],
  "temperature": 0.7,
  "max_tokens": 1024
}

200 OK returns the standard OpenAI shape: id, object: "chat.completion", choices[], usage{}. With stream: true, you receive text/event-stream chunks (data: {…}) terminated by data: [DONE].

`GET /v1/models`

Returns the single supported model.

{
  "object": "list",
  "data": [
    {
      "id": "deepseek-v3.2",
      "object": "model",
      "owned_by": "deepseek"
    }
  ]
}

Rate limits

Default: 60 requests per minute per API key. Exceeding this returns 429 rate_limit_exceeded. Bump the cap by emailing support — we'll lift verified accounts to 600 rpm.

What we don't expose

embeddings, images, audio — DeepSeek V3.2 is text-only.
Function calling / tool use — supported by upstream; passthrough lands once Cursor ships first-class support.
File uploads, threads, assistants — out of scope at MVP.

← previousQuickstart

next →Errors