Platform overview

Token Spark brings together model and compute providers through one API, with agent services and options for compute providers.

What Token Spark does

Token Spark connects individuals and teams to AI services. A single platform API key lets you call available models within your account permissions, without integrating with each upstream provider separately.

  • Integration: brings model and compute services into a shared routing and metering system.
  • Distribution: makes model services available through a common API.
  • Additional services: offers agents and model hosting for different use cases.

Core features

  • Common API: OpenAI-compatible request formats, subject to each model’s supported features.
  • Multiple models: browse the marketplace for currently available text, image and other models.
  • Routing: allocates requests using your strategy, resource availability, cost and performance.
  • Failover: some failures can trigger retries or a resource switch; successful completion is not guaranteed.
  • Provide compute: eligible resources can apply for listing, with earnings settled under the applicable agreement.
  • · The API uses OpenAI-compatible request formats, but supported parameters and features vary by model, upstream service and client. Test your chosen model first; do not assume every OpenAI feature is supported.

Get started

1

Create an account

Register and sign in to the dashboard.

2

Create an API key

Open API key management, create a key and keep it private.

3

Check your balance

Visit balance and top-ups. Payment methods, currency and amounts depend on the actual checkout.

4

Test a request

Follow your first API call, choose an available model and check the usage record.

Community and support

Scan the QR codes to follow the WeChat official account or join the Telegram community for updates and discussion.

WeChat official accountWeChat official account
Telegram communityTelegram community

Docs center

The docs center covers quickstart, API guides, third-party app integration, business and billing notes, and an error-code reference. All examples follow the OpenAI-compatible protocol.

The fastest path: register, create a key in the console, and point your base_url at https://tokenspark.uno/v1. The rest of your code stays the same.

If you run into trouble during integration, check the error-code reference and FAQ first; if the issue remains, reach the platform through in-site tickets.

FAQ

How much code do I need to change?

Usually one line: point base_url at https://tokenspark.uno/v1 and fill in your platform key, then set the model name as needed.

Which third-party apps are supported?

The third-party apps section of the docs center lists integration guides for verified clients and tools, including coding assistants such as Claude Code.

How do I reconcile billing?

The console keeps a per-call log with model, token usage, unit price and settled amount — every line is traceable.