LLM API Reference

​

Introduction

​

Authentication

• API keys should be provided via HTTP Bearer authentication:
  Authorization: Bearer GMI_API_KEY
  
• Never expose API keys in client-side code
• Load keys from environment variables or key management services
• For multi-organization access, specify headers:
  curl https://api.gmi-serving.com/v1/models \
    -H "Authorization: Bearer $GMI_API_KEY" \
    -H "X-Organization-ID: your_org_id"
  

​

List Models

​

Example Request

curl https://api.gmi-serving.com/v1/models \
  -H "Authorization: Bearer $GMI_API_KEY"

​

Response

{
  "object": "list",
  "data": [
    {
      "id": "",
      "object": "deepseek-ai/DeepSeek-R1",
      "created": 1687530000,
      "owned_by": "public",
    },
    // ... other models ...
  ]
}

​

Response Parameters

​

Create Chat Completion

​

Authorization

Authorization: Bearer <token>

​

Request Body

{
  "model": "deepseek-ai/DeepSeek-R1",
  "messages": [
    {
      "role": "user",
      "content": "Hello!"
    }
  ],
  "max_tokens": 2000,
  "temperature": 1
}

​

Parameters

​

Response

{
  "id": "chatcmpl-123",
  "object": "chat.completion",
  "created": 1677652288,
  "model": "deepseek-ai/DeepSeek-R1",
  "choices": [
    {
      "message": {
        "role": "assistant",
        "content": "Hello! How can I help you today?"
      }
    }
  ],
  "usage": {
    "prompt_tokens": 9,
    "completion_tokens": 12,
    "total_tokens": 21
  }
}

​

Response Fields

​

Important Notes

1. Use response_format: {"type": "json_object"} for JSON mode
2. Streaming responses include usage stats in final chunk
3. Default context handling differs from Other provider (truncates instead of erroring)
4. Multiple penalties interact - use carefully to avoid quality degradation

​

Key Notes

1. Parameter support varies by model - check model documentation
2. New projects should use Responses format for latest features
3. Organization/project usage is tracked via headers
4. Find organization/project IDs in settings pages