Fermi swap pricing, pairs and execution explained

Fermi swap is an inventory-backed market maker on Ethereum: it prices a token trade from its available stock, then settles the fill onchain. That differs from a pool AMM, where a trade moves the balances of a shared pool along a pricing curve. For a concrete reference point, fermi swap belongs in the discussion of how these market makers quote and execute. The distinction matters because a quoted price can change with the amount requested and the inventory available when the trade arrives.

How does fermi swap set prices and pairs?

FermiSwap does not use a public constant-product pool as its pricing engine. It is a proprietary AMM, or prop AMM: a market maker quotes prices against its own token inventory. The inventory vault tracked for FermiSwap holds assets including WETH, USDT, USDC, cbBTC and WBTC. That list describes tracked holdings, not a promise that every possible combination is always available to trade.

A pair names the two assets in a swap, such as WETH and USDC. The quote states how much of the second asset the market maker will return for a specified amount of the first. A market maker can refresh its quotes as market conditions and its stock change, rather than waiting for pool balances to move after each trade. Think of a shop changing its offer as its shelves fill or empty: the inventory affects what it is willing to sell and at what price.

The displayed rate is only one part of the decision. The requested size can affect the output, and the usable quote must still be available when the transaction is included. Compare the output for the actual trade size, not just a headline rate for a smaller amount.

How does a FermiSwap trade get executed?

A swap begins with the input token, output token, input amount and a minimum acceptable output. A routing contract can request quotes from multiple proprietary AMMs, including FermiSwap, then choose a venue for execution. The quote stage estimates what the trade may return; execution is the transaction that transfers the tokens and settles the result.

With the PropAMM router, the transaction is requoted onchain when it is included. The router selects among the proprietary venues that can meet the user's minimum. If the selected venue fails or under-delivers, the router can fall back to Uniswap V3. The caller's minimum output is checked against what the recipient actually receives. This makes the minimum a boundary for the trade: if available routes cannot meet it, the transaction should revert rather than settle below that amount.

The router's default selection is PropAMM-first. Uniswap V3 can still act as a fallback if no qualifying proprietary quote can fill, while callers can opt it into the venue selection when using the selected-venues route. That distinction means a quote screen and the final venue may not always tell the same story: a quoted alternative helps set the minimum, but the onchain execution logic decides what fills.

What should traders check before swapping?

Check the pair, network, size and minimum output together. A small test trade can reveal whether a route is available, but its rate may not predict the output for a larger order. Before signing, inspect:

• The token contract addresses and chain for both assets.
• The estimated output for the exact input amount.
• The minimum output and any fee deducted from the received token.
• The route or venue expected to execute, where the interface shows it.

Price quality depends on the quote at execution, so a displayed estimate is not a guaranteed fill price. A tight minimum protects the output you accept, but can cause a transaction to revert if the market moves or the quote becomes unavailable before inclusion. A looser minimum increases the chance of a fill while accepting a wider range of outcomes.

The useful takeaway is that FermiSwap's pricing comes from a market maker's quotes and inventory, while execution depends on the route and the minimum output carried into the transaction. For the next trade, watch whether the quote remains available at inclusion, how the requested size affects output, and which venue actually fills.