A BETTER DEAL FOR YOUR AI STACK

Better tokens.
Lower costs.

More room to build. Less spent on tokens. One familiar API for the models you choose, built around better pricing and clear usage.

Explore the API See how it works

In development. A first look at what we’re building.

YOUR STACK, CONNECTEDPREVIEW

ONE GATEWAY. FAMILIAR PROTOCOLS.

OpenAIAnthropic

api.tokendeal.io

01 Familiar API formats

02 Your choice of model

03 Costs you can understand

01 / THE IDEA

Spend less on access.
Do more with AI.

TokenDeal brings model access and usage into one place, so your next idea doesn’t need another integration.

A familiar place to start

Better value, model by model.

Discounted access is the goal. Published offers will show the actual model, its rates, and the basis for any savings claim.

Your model. Your decision.

Choose the model that fits your work. Compatibility and supported features will be clear before you send a request.

Know where your tokens go.

Clear usage, separate input and output costs, and explicit cache accounting where the model supports it.

02 / ONE CONNECTION

New gateway.
Familiar syntax.

OpenAI-compatible and Anthropic-compatible interfaces. The same planned TokenDeal key, with the right base URL for your client.

OPENAI CLIENT BASE URL

https://api.tokendeal.io/v1

POST/v1/chat/completions

Keep /v1 in the client base URL. Streaming, tools, and other features depend on the selected model.

Configuration preview. Public API access is not available yet.

example.mjs

import OpenAI from "openai";

const client = new OpenAI({
  apiKey: process.env.TOKENDEAL_API_KEY,
  baseURL: "https://api.tokendeal.io/v1",
  maxRetries: 0,
});

const response = await client.chat.completions.create({
  model: process.env.TOKENDEAL_MODEL,
  messages: [{ role: "user", content: "Hello, TokenDeal." }],
});

ANTHROPIC CLIENT BASE URL

https://api.tokendeal.io

POST/v1/messages

Use the host as the base URL. The SDK adds /v1/messages. Claude Code compatibility will be verified per model.

Configuration preview. Public API access is not available yet.

example.mjs

import Anthropic from "@anthropic-ai/sdk";

const client = new Anthropic({
  apiKey: process.env.TOKENDEAL_API_KEY,
  baseURL: "https://api.tokendeal.io",
  maxRetries: 0,
});

const response = await client.messages.create({
  model: process.env.TOKENDEAL_MODEL,
  max_tokens: 1024,
  messages: [{ role: "user", content: "Hello, TokenDeal." }],
});

Set TOKENDEAL_API_KEY and TOKENDEAL_MODEL when access launches. Published models and capabilities will determine which requests are supported.

Enable JavaScript to switch examples and use copy buttons. The Anthropic client base URL is https://api.tokendeal.io; its request path is /v1/messages.

A place for your existing clients.

Planned endpoints

Planned TokenDeal endpoints. All public API access is in development.
Method & path	Made for	Release scope
POST`/v1/chat/completions`	OpenAI-compatible chat	Core launch
POST`/v1/messages`	Anthropic-compatible messages	Core launch
GET`/v1/models`	Model and capability discovery	Core launch
POST`/v1/responses`	Responses API workflows	After compatibility checks
POST`/v1/images/generations`	Image generation Alias: `/v1/images`	Planned extension
POST`/v1/audio/speech`	Text to speech	Planned extension
POST`/v1/embeddings`	Embedding vectors	Planned extension
POST`/v1/rerank`	Document reranking	Planned extension

03 / YOUR PACE. YOUR PLAN.

Start with what you need.
Build from there.

Experiment as you go, or make room for a regular workflow. Two ways to fund your requests are on the way.

Planned

FLEXIBLE BY DESIGN

Pay as you go.

For new ideas and workloads that change.

Add prepaid credit to your wallet.
Pay for usage at the published model rates.
Keep spending limits in your control.

Wallet balanceUsage-based

Planned

ROOM FOR YOUR ROUTINE

Monthly Token Plans.

For the work you come back to every month.

A monthly allowance for eligible models.
Your allowance resets each month.
Optional PAYG fallback, only if you enable it.

Plan allowanceMonthly reset

Wallet credit and plan allowance stay separate. Prices, eligible models, allowance units, and renewal terms will be published before plans become available.

04 / MEET TOKENDEAL CODE

From your first prompt
to your next commit.

A coding companion for your terminal and editor, built on OpenCode. We’re bringing the same model choice and cost clarity closer to where you write code.

In development

tokendeal codePREVIEW

Build something
worth shipping.

What are we building today?

YOUR PROJECTYOUR MODEL

A FEW GOOD QUESTIONS

Before you
plug in.

The important details, without the fine-print hunt.

Can I use TokenDeal today?

This is an early preview of TokenDeal. Public API access, accounts, billing, and TokenDeal Code are still in development. You can explore the planned API formats here; there is nothing to purchase yet.

Which models will be available?

The model catalog is being prepared. Each published model will list its supported endpoints, features, availability, and rates. We’ll show verified offers rather than promise a model before it is ready.

Will my existing SDK work?

TokenDeal is being built for OpenAI-compatible and Anthropic-compatible clients. The OpenAI base URL includes /v1; the Anthropic base URL uses the host only. Features such as tools, vision, and streaming will be verified for each model. The API examples above show the planned configuration.

How much can I save?

Lower-cost model access is central to TokenDeal. Exact savings will depend on the model and offer. We’ll publish input and output rates, applicable cache rates, and the reference behind each discount when verified offers are available.

How do Token Plans work with PAYG?

Token Plans will provide an allowance that resets monthly, kept separate from prepaid wallet credit. PAYG fallback will be optional. If you enable it and a request cannot be covered by your plan, the whole request will use your wallet, subject to your spending limits. Pricing and detailed plan terms will be published before launch.

BETTER VALUE STARTS HERE

Your next idea.
A better token deal.

Explore the API

Better tokens.Lower costs.

Spend less on access.Do more with AI.