API Reference Overview

Conventions shared by every Calgine Tokens API: authentication, endpoint groups, error codes and pagination.

Calgine Tokens offers OpenAI-compatible and Anthropic-compatible chat endpoints, plus helper endpoints such as the model list and the unified async task API. All endpoints share one set of authentication and error conventions.

Basics

Item	Value
Base URL	https://api.calgine.com.cn
Authentication	Header Authorization: Bearer sk-your-token
Request body	JSON, Content-Type: application/json

Authentication

Every endpoint expects your API token in the request header:

curl https://api.calgine.com.cn/v1/models \
  -H "Authorization: Bearer sk-your-token"

• Create API tokens under Workspace → API Tokens in the Calgine Tokens Console
• The Anthropic-compatible endpoint also accepts the x-api-key header (matching the official SDK behaviour)
• Management endpoints (/api/...) use an Admin Key instead of an API token — see What is an Admin Key

WARNING

Never embed tokens in frontend code or public repositories. If a token leaks, delete it in the console and create a new one immediately.

Endpoint groups

Chat endpoints

Endpoint	Protocol	Description
POST /v1/chat/completions	OpenAI-compatible	Chat completions with streaming and tool calling — see Language Models
POST /v1/responses	OpenAI-compatible	Responses API, OpenAI models only
POST /anthropic/v1/messages	Anthropic-compatible	Messages API for Anthropic SDKs — see Language Models

Model list

Endpoint	Description
GET /v1/models	Returns all model IDs available to the token's group, in OpenAI-compatible format

{
  "object": "list",
  "data": [
    { "id": "gpt-4o", "object": "model", "created": 1711234567, "owned_by": "openai" }
  ]
}

Async task endpoints

Long-running generation tasks (video, image, music) go through the unified async API — see Async Tasks for details:

Endpoint	Description
POST /v1/task/submit	Submit a task and get a task ID immediately
GET /v1/task/{task_id}	Query task status and results

Error codes

Failed requests return a non-2xx status code with an OpenAI-style error object:

{
  "error": {
    "message": "Invalid API key provided",
    "type": "authentication_error",
    "code": "invalid_api_key"
  }
}

Status	Common causes
400	Missing or invalid parameters, model unavailable, malformed request body
401	Invalid, deleted or unauthorised token
402	Insufficient balance — top up in the console
404	Wrong endpoint path or task not found
429	Rate limited — slow down and retry
5xx	Service temporarily unavailable or upstream timeout — retry later

An async task that is accepted but fails during execution does not produce an error status code; instead the task query returns status: "failed" with the reason in the error field — see Task statuses.

Pagination

List endpoints (such as the Admin Key log and key lists) use page-based pagination:

Parameter	Description
p	Page number, starting at 1
page_size	Items per page; each endpoint has its own upper limit

Responses include total (total items), page and page_size; some endpoints also provide has_more to indicate whether another page follows. Over wide time ranges total may come back as -1 for "unknown" — in that case page by has_more instead. See the Admin Key API Reference for details.

Next steps

• Quickstart: make your first request in 5 minutes
• Language Models: calling examples for the OpenAI, Anthropic and Gemini protocols
• Async Tasks: the full submit, poll and callback workflow
• Admin Key API Reference: eleven endpoints for managing your account from scripts