ChineseModels.ai

About ChineseModels.ai

One API for Chinese AI models — compare them, integrate them and ship, without maintaining a separate relationship with every provider.

What we provide

ChineseModels.ai is an access layer, not a model vendor. We route requests to the providers behind each model and keep the commercial and operational details in one place, which leaves your team free to focus on the product.

A unified API surface

One set of credentials and one base URL in front of models from several Chinese providers. Requests may use the OpenAI Chat Completions, Anthropic Messages or Gemini generateContent shape, so existing client code usually works after changing the endpoint, the key and the model ID.

Flexible model access

Chat, reasoning, vision and video models sit behind the same billing account. Switching between them is a change to the model field rather than a new contract, which makes it practical to compare quality and cost on your own inputs before committing.

Controlled access and usage

Keys are scoped per account and can be named for each environment, then rotated or revoked independently. Usage, spend and per-request outcomes are recorded in the console, so cost and behaviour are visible without building your own metering.

Getting started

  1. Create an account. Sign up with an email address. No payment details are needed to look around the catalogue.
  2. Add credit. Top up from the console. Balance is shared across every model on the platform.
  3. Create an API key. Generate a key in Console → API Keys and store it in your secret manager, not in source control.
  4. Pick a model and send a request. Choose a listing in Model Square, copy its model ID and call the compatible endpoint with the key you created.
  5. Watch usage. Check request-level logs and remaining balance as traffic grows, and set limits before you go live.

Use AI responsibly

Access is conditional on using these models lawfully and with care for the people affected by the output. Our Usage Policy sets out prohibited content and conduct, and upstream providers’ own terms apply in addition to ours. Applications that make consequential decisions about people — credit, employment, housing, health, legal outcomes — need human review before any decision is acted on.

Generated output can be wrong, out of date or biased, and should be treated as a draft rather than a fact. If you redistribute model output publicly, label it as AI-generated and keep a way for people to report harmful results to you.

Operator and contact

Support hours: Monday to Friday, 09:00–18:00 UTC. We aim to answer account and integration questions within one business day.