Introducing Token$_ — one API for every model, now in private beta

Token$_ (tokens.co.jp) is an AI API platform based in Japan. You point your existing SDK at api.tokens.co.jp and get the models you already use — the Claude family, GPT, Gemini, DeepSeek, plus image and video models — through one account, one balance, and one dashboard.

Why another API platform?

Teams we work with run more than one model, and every provider adds its own billing, quota management, and failure modes. The pattern we kept seeing: engineering time going into plumbing that has nothing to do with the product. An aggregation layer solves boring problems — one contract, one prepaid balance, automatic failover when an upstream has a bad day.

We're not the first to build one. We are building this one for teams in Japan: contracts and support that work here, and an API that behaves exactly like upstream.

What works today

• Anthropic-native endpoint. Full /v1/messages support — streaming, tool use, long context — for Claude models.
• OpenAI-compatible endpoint. /v1/chat/completions for every language model in the catalog. Switch the base URL and keep your code.
• Prepaid credits. Billed on exact token usage. Your dashboard shows balance and per-key usage in real time.
• Automatic failover. Multi-channel upstream routing reroutes around outages in under a second.

The full list is on the Models page, and it grows with customer demand.

How the private beta works

There's no self-serve signup. Tell us which models you run and your expected monthly volume through the contact form, and we set up your workspace and API keys by hand — usually within one business day. Onboarding a limited number of teams at a time keeps support honest.

If you're running a real AI workload and want it behind one API, we'd like to hear from you.

Token$_ is an independent platform and is not affiliated with Anthropic, OpenAI, Google, or any other model provider.