Official frontier AI models for a fraction of the list price.

How it works

1. Create an account

   Register and get 10 free requests to get started and test everything out.

2. Create your first key

   Create a key for each provider pool you need.

3. Point your client at Token Supply

   Change the base URL and API key in your tool of choice, eg. Codex, Claude Code, OpenCode, an SDK, Hermes and more.

4. Add credit and keep going

   After the free requests, usage debits the prepaid wallet. Requests stop when the balance reaches zero.

Client setup

• Use Codex with a custom API endpoint

  Keep the Codex CLI, app, or VS Code extension you already use. Point the provider at Token Supply, set a plan key, and keep the Responses API.

• Use Claude Code with a custom API endpoint

  Keep Claude Code CLI or the VS Code extension. Replace the Anthropic base URL and auth token with Token Supply, then set reasoning effort.

• OpenCode custom provider configuration

  Keep OpenCode's OpenAI, Anthropic, Gemini, or Grok preset. Replace baseURL and apiKey, set context limits, and use variants for reasoning effort.

• Configure Cursor with an OpenAI-compatible API

  Keep Cursor's OpenAI preset. Paste a Token Supply key, turn on Override OpenAI Base URL, and choose a model from that key's plan.

• Use the OpenAI SDK with a custom base URL

  Keep the OpenAI SDK, curl, or any OpenAI-compatible client. Change the API key and base URL. The model must belong to the plan on that key.

• Use the Anthropic SDK with a custom base URL

  Keep the Anthropic SDK. Change the API key and base URL. Claude Code uses the same host; do not add /v1 to ANTHROPIC_BASE_URL.

• Configure Hermes with an OpenAI-compatible API

  Keep Hermes's custom endpoint setup. Replace the base URL, set the context window, and store the Token Supply key in .env.

• Write client config files from the terminal

  A setup CLI is coming soon. Until then, create a key in API Keys and paste the values from the matching client guide.

Costs and usage

• AI API token cost calculator

  Price a coding-agent workload here. The same calculator stays on the homepage. Change the model and monthly token mix, then keep the same client.

• Reduce coding-agent API costs

  Most coding-agent spend is model tokens. Keep Codex, Claude Code, OpenCode, or your SDK, compare listed rates with public list prices, and put a cap on the wallet.

• Track AI API token usage

  Prepaid billing only helps if you can see the debit. Token Supply records usage per request, rolls it up in the dashboard, and stops new calls when the wallet hits zero.

Flagship model prices

Listed rates per 1M tokens as of 2026-09-28. Live rates load on the page.

• GPT-6 Sol (gpt-6-sol): $0.60 input / $3.00 output per 1M tokens on Token Supply, vs $2.00 / $10.00 official. 70% less on input.
• Claude Opus 5.5 (claude-opus-5-5): $1.80 input / $9.00 output per 1M tokens on Token Supply, vs $4.00 / $20.00 official. 55% less on input.
• GPT-5.6 Sol (gpt-5.6-sol): $1.50 input / $9.00 output per 1M tokens on Token Supply, vs $5.00 / $30.00 official. 70% less on input.
• Grok 4.7 (grok-4.7): $0.60 input / $1.80 output per 1M tokens on Token Supply, vs $2.00 / $6.00 official. 70% less on input.
• Gemini 3.1 Pro (gemini-3.1-pro): $0.60 input / $3.60 output per 1M tokens on Token Supply, vs $2.00 / $12.00 official. 70% less on input.

All model prices

Official models

Only real frontier models, same IDs you already use. Not cheaper substitutes.

Published rates

Don't take our word for it. Listed and official rates for every model are published and dated, so you can check the gap yourself.

Listed vs official rates

Try it yourself

New users get 10 free requests to get started and test everything out before adding credit.

Start free — 10 requests

Details

Do I need to change my agent or IDE?

No. Simply change your agent's config file to use Token Supply's API URL and your TS key. It's easy to set up.

What does Token Supply provide?

Cheaper API access to the frontier models you already use. No cheaper substitutes. Your existing agent provides the workflow.

Which clients work?

Codex, Claude Code, OpenCode, the OpenAI SDK, the Anthropic SDK, and curl.

How do I get started?

Create an account and you get 10 free requests to test everything out. No credit card / wallet top up required.

Related

• AI API token cost calculator — Compare listed Token Supply rates with public list prices.
• Reduce coding-agent API costs — Keep your client and lower token spend on the same workload.
• Track AI API token usage — Prepaid wallet, usage dashboard, and per-key spend.
• Codex custom endpoint — config.toml, auth.json, and the Responses API.
• Claude Code custom endpoint — ANTHROPIC_BASE_URL, auth token, and settings.json.