FAQ

General

What is TOKI?

TOKI provides model API services for developers and supports calling available platform models through OpenAI-compatible endpoints.

Is TOKI compatible with OpenAI?

Yes. TOKI model APIs follow common OpenAI API request and response formats. In most cases, you only need to change baseURL and the API key in your OpenAI SDK setup.

What Base URL should I use?

Documentation examples use:

https://www.tokiai.ai/v1

If the console, operations team, or enterprise onboarding material provides a dedicated endpoint, use that endpoint instead.

Which models are supported?

Use the Models page and console as the source of truth. Model availability, pricing, context length, and capabilities can change by supplier and model version.

Token Plan is the Token package product provided by the platform. After purchase, quota can be used for supported model usage scenarios. Token amount, price, validity period, and limitations are determined by the product center.

How do I check Token usage?

Check wallet, Token balance, or related usage pages in the console. Actual usage depends on model, input context, output length, and conversation history.

Is there a free tier?

This documentation does not promise a fixed free quota. Trial quota, granted quota, or campaign packages are subject to the console and campaign rules.

Technical

Does TOKI support streaming?

Yes. Set stream: true when calling /v1/chat/completions, then read incremental data from the Server-Sent Events stream.

Are tool calls or JSON mode supported?

The parameter shape follows OpenAI-compatible conventions, but specific capabilities depend on the selected model and current server support. Validate the smallest request shape with the target model first.

How should I handle failed requests?

Read the HTTP status code and error.message in the JSON error body. Common causes include invalid API key, disabled key, insufficient quota, unavailable model, unsupported model parameter, or overly frequent requests.

FAQ