Documentation

Inference Markets leverages a new primitive: Intelligence. Every market launched here is quoted in a unit of model output instead of a stablecoin, its holders are paid the trading fees in…

An inference coin is an ERC-20 token that stands for one unit of machine work: a million frontier model tokens, a billion open-weight tokens, an hour of an agent.

The inference index is the price of intelligence, written on-chain. Once an hour a Chainlink Functions network reads OpenRouter's public list of model prices, works out what the top…

DIEM is Venice AI's compute token. One DIEM, staked in a Venice account, gives that account one dollar of Venice API credit every day, in perpetuity, for as long as the stake is held.

Coinbase issues tokenized stocks on Base under the B20 standard. Six of them cover the companies building the intelligence: NVIDIA, Alphabet, Meta, Microsoft, Amazon and SanDisk.

Launching is one form and one transaction. You give the token a name, a symbol and an image, pick what it is paired with, choose the trading fee, and make the first buy.

Every market on Inference Markets holds its entire supply as liquidity from the block that created it.

Every trade goes through the LaunchRouter. You pay in ETH, USDC, or the pair coin you already hold, and the router crosses whatever pools sit between your asset and the market and settles…

Every swap in a market pays the LP fee its creator chose at launch, between 1% and 3%. The fee accrues to the two liquidity positions the launchpad owns until someone calls collectFees,…

Holding a market token pays you a share of that market's trading fees every block. The fees accrue to your balance inside the token itself, and they are yours the moment you claim.

Every inference coin can be burned for what it says it is worth. Send INT to the RedemptionVault and it burns your coin and pays you its index value in the same transaction — in USDC, or in…

Coverage is what the vault holds for a coin divided by what that coin owes its holders, published on-chain as a number anyone can read.

The protocol token is the venue's official token, and it stands on its own: it is a plain fixed-supply ERC-20 that can be deployed and launched anywhere, on this venue or another, before or…

Nothing in Inference Markets needs an operator, but some things need someone to call them: a fresh index sample has to be requested and published, a peg has to be repriced when the feed…

Inference Markets is one set of Solidity contracts, built with Foundry on Solidity 0.8.26 and deployed to Base.

Read this before you trade or redeem. It describes how Inference Markets works today, and nothing here is softened.