Introduction

CanalAPI is an AI model API aggregation platform that exposes a single, OpenAI-compatible endpoint for hundreds of large language models — including overseas providers (OpenAI, Anthropic, Google) and Chinese providers (DeepSeek, Moonshot/Kimi, Zhipu, Alibaba/Qwen and more).

It is operated by a Japan-incorporated entity and serves two complementary audiences:

• Developers and teams inside China that need reliable access to overseas frontier models, with local payment methods (Alipay / WeChat Pay / Stripe) and compliant Japanese invoices.
• Developers in Japan and the rest of the world that want a fast, affordable gateway into Chinese open-weight and proprietary models.

What this site covers

This site is the public documentation for CanalAPI. It is intended for:

• End users who manage an account, API keys, and billing.
• External developers who want to integrate CanalAPI into their own applications.
• Teams evaluating CanalAPI for production workloads.

It is not a reference for CanalAPI's internal services or operations.

Highlights

• One endpoint, many models. A single Base URL exposes hundreds of models. Pick the model name in the request body — no per-provider SDK required.
• OpenAI SDK compatible. Chat Completions, embeddings, streaming and function calling work with the upstream OpenAI SDKs as drop-in clients.
• Transparent metering. Every request is recorded with token counts and cost. Real-time balance and usage logs are available in the console.
• Production-friendly. Per-key rate limits, scoped permissions, soft and hard quotas, and clear error semantics.

Where to go next

• Getting Started — create an account, get an API key, send your first request.
• Developer Guide — authentication, errors, streaming, models, rate limits.
• Examples — curl, JavaScript, Python.
• FAQ — common questions.

Note: CanalAPI is under active development. Some endpoints and features may evolve. Breaking changes are announced in the Changelog.