Skip to content

About

Docs, guides, repos and live endpoints for Inferit, the peer-to-peer market for AI inference. Testnet demo.

Topics

Resources

Stars

0 stars

Watchers

0 watching

Forks

Latest commit

 

History

2 Commits

Folders and files

NameName
Last commit message
Last commit date
 
 
 
 
 
 

Repository files navigation

Awesome Inferit

Website Docs X License: CC0-1.0 Links

Docs, guides, repositories and live endpoints for Inferit, in one index. Inferit is a peer-to-peer market for AI inference: anyone can sell inference for any model, at their own price, and buyers and agents pay per request through one OpenAI-compatible API.

Status: Testnet demo on Whitechain Sepolia. Prices settle in Inferit Test Credit (ITC), a test token with no monetary value. The contracts are not audited. Inferit has no token.

Contents

Start here

  • inferit.cc: what Inferit is, with each live model's price against its list price (list prices from an OpenRouter snapshot, 2026-10-02).
  • inferit.app: the app. Buy holds your balance, ceiling, API keys and usage; Sell holds your machines, offers and prices.
  • Documentation: the overview, units (µ, one micro-dollar) and the two ways to pay.
  • Quickstart: from zero to a settled request on Whitechain Sepolia, in the app or by hand.

Buy inference

  • API reference: every endpoint, authentication with an API key or x402, money in µ, errors and response headers.
  • Automatic model routing: how "model": "auto" picks a model. Its policies, its hints, and the quality ranking it reads (Artificial Analysis Intelligence Index, read 2026-10-06).
  • Wallet and keys: one balance, a key per teammate or agent with its own spend limit, the ceiling and its expiry, usage and CSV export.
  • Models with live offers: the catalog, list prices and the lowest live all-in prices.

The API is OpenAI-compatible. Point any OpenAI client at https://api-production-c74b9.up.railway.app/v1 with your key, and name a model with a live offer or send "model": "auto". Auto chooses among ranked models that have a healthy seller; when none has one, it answers 503 no_eligible_model and nothing is charged.

Agents and x402

  • Inferit for agents: how an agent buys inference per request, with no account and no API key.
  • agents.md: the same guide for machines: model auto, x402 v2 pay per request, limits, errors and the ITC faucet.
  • llms.txt and llms-full.txt: the site and docs for language models. Every page on inferit.cc also has a markdown twin: add .md to its path.
  • x402 discovery: what a client signs for: network, asset, payee and the minimum amount (10 000 µ, that is 0.01 ITC).

A call with no key gets 402 Payment Required with a quote. The agent signs an EIP-3009 authorization, resends the same body with the PAYMENT-SIGNATURE header, and is charged the metered cost; the rest stays its escrow balance.

Framework and tool guides

  • LangChain: Python and JavaScript, model auto or a named model, and x402 from LangChain JS.
  • Vercel AI SDK: Inferit's base URL and key in the AI SDK's OpenAI-compatible setup, and x402 through a custom fetch.
  • LiteLLM: the SDK call with the openai/ prefix, and a proxy config.
  • CrewAI: Inferit as the model of CrewAI agents.
  • OpenCode: Inferit as a custom OpenAI-compatible endpoint in OpenCode, with model auto or a named model.
  • Cursor: Inferit models through Cursor's OpenAI base URL override.

Sell inference

  • Sell from your own PC in 10 minutes: run a model in Ollama or LM Studio, connect it with one command, set a price and list it. The 10 minutes assume the model is already downloaded and your wallet is registered as a seller.
  • Selling inference: licences, registration, health checks, availability and payouts to your own wallet (test ITC on the testnet).
  • inferit-seller: the open-source seller app (Apache-2.0, on npm). It finds Ollama, LM Studio, vLLM or llama.cpp and dials out to the market, so you need no open port and no domain.
npx -y inferit-seller connect --token icn_… && npx -y inferit-seller install-service

The token comes from Add a machine in the Sell workspace. Buyers pay your price plus a 5% platform fee, and you are paid per request to the payout wallet you register.

Payments on Whitechain

Inferit runs both facilitators: the one inside the market API and the open-source one, which the market API does not use.

Live API endpoints

Base URL: https://api-production-c74b9.up.railway.app

Endpoint What you get
GET /health The network, the MarketEscrow contract, and the state of the settlement and x402 keys
GET /v1/models The model auto rows, then the catalog with list prices and the lowest live all-in prices
GET /v1/routing The auto policies, the hints, the quality ranking and the models auto can choose from now
GET /.well-known/x402 The x402 requirement a client signs for
GET /metrics Settlement and chain figures, including the settlement cadence
GET /llms.txt The API's own summary for language models
POST /v1/chat/completions Chat completions, streaming included, with an API key or x402

Repositories

Repository What it is
inferit-seller Sell inference from your own computer with one command. Apache-2.0, on npm
whitechain-x402-facilitator The first public x402 v2 facilitator for Whitechain, live on Whitechain Sepolia (testnet). Apache-2.0
inferit-examples Runnable examples for LangChain, the Vercel AI SDK, LiteLLM, CrewAI and OpenCode, paying with an API key or x402
inferit-agent-example A minimal TypeScript agent that pays Inferit per request with x402: no account and no API key
model-catalog The models the Inferit API knows, as open data: ids, aliases, context windows and licence classes (CC BY 4.0)
awesome-x402-whitechain x402 resources for Whitechain: the spec, SDKs, EIP-3009, the testnet, facilitators and examples
branding The Inferit logo files, colours, type and usage rules

Inferit runs the market API and the router; their code is private.

Security and trust

  • Security & trust: what MarketEscrow is designed to enforce, what the API keeps private, what is public on chain, and what is not verified yet.
  • Security policy: how to report a vulnerability, what is in scope, and the terms for good-faith research.

The API stores token counts and costs, never prompts or completions. The seller that serves a request does see it.

Community

Roadmap

  • Inferit roadmap - What is live, what is next and what is planned, one issue per item.

Contributing

Add one line per resource, in the section it belongs to, with a link that resolves and a description that says what it does. Only public resources that work with Inferit today. Open a pull request; the link check runs on it. By contributing, you dedicate your contribution to the public domain under CC0 1.0.

License

CC0 1.0 Universal: the list is dedicated to the public domain. The Inferit name and logo are not covered: see InferitAI/branding.

About

Docs, guides, repos and live endpoints for Inferit, the peer-to-peer market for AI inference. Testnet demo.

Topics

Resources

Stars

0 stars

Watchers

0 watching

Forks

Releases

Packages

Contributors