Same models. Compare the cost.
Loading the shared-model usage comparison…
One OpenAI-compatible API. Provider cost + 5%. Start with free models, or top up from $0.001 in USDG on Robinhood Chain or USDC on Base — one gasless signature.
$0 models need no key and no payment. Start with the free chat lane.
Loading the frontier catalog comparison…
Loading prices and context windows…
No account. No KYC. Credits never expire. The comparison includes models where mini costs less.
The headline compares modeled token costs for the same observed prompt and output usage on shared models. Below, compare each published input and output rate. Prices include our markup.
Loading prices…
| Model | Our in / out USD per M tokens | Mini in / out USD per M tokens | Input savings | Output savings | 3:1 blend savings |
|---|---|---|---|---|---|
| Fetching the published prices… | |||||
Savings = (mini price − our price) ÷ mini price. The final column assumes 3 input tokens per output token; your mix may differ. A free mini rate has no percentage denominator.
Published catalog entries absent from mini's catalog, with their input/output rates and context windows.
loading the live catalog…
Side by side, as it is right now — recomputed live from both published catalogs and yesterday's real usage. Every model at x402 door cost + up to 5%; their +5% named / +8% auto on top of a $0.50 deposit.
Shared-model token prices weighted by ecosystem request counts, assuming a 3:1 input/output mix. This is a separate comparison from the headline's observed token mix.
Based on the same observed shared-model prompt and output usage as the headline.
See the published rates and context windows above. Model availability comes from the current catalog comparison.
A tenth of a cent. Gasless x402 top-up in USDG or USDC — you sign a payment authorization.
Choose a named model or an auto tier. The router selects an upstream for each request; published input and output rates include our 5% markup.
Use Chat Completions, Responses, or Anthropic Messages with one base URL and API key. Tool calls and results pass through the SDK adapters; support depends on your chosen model.
SillyTavern and every OpenAI-compatible client. Or our own chat with a free try-it lane.
OpenAI SDK Chat Completions in Python, Node, or Go. Set the base URL, API key, and model ID.
Skip the key: sign the quoted price and send an X-PAYMENT header. The quote includes our 5% markup and settles in USDG or USDC before the call; the facilitator pays gas. Use a topped-up key for billing from measured token usage.
Point an OpenAI Chat Completions client at the endpoint.
OPENAI_BASE_URL=https://infinitesimal-router.fly.dev/v1
OPENAI_API_KEY=ir-live-…
OPENAI_MODEL=openai/gpt-6.1-sol-pro
Base URL · API key · Model ID — same three lines as anywhere else.
Top up in USDG on Robinhood Chain or USDC on Base. No transaction to send, no gas to hold: the wallet signs one EIP-3009 message and the facilitator relays. x402 under the hood, invisible under the same experience.
Pay in USDG — Global Dollar. Robinhood Chain settlements are tracked in the relay ledger; automatic bridging to Base is not currently implemented.
Native Circle USDC, EIP-3009. The flagship x402 chain.