Memecoin terminal take rates: what Terminal, Trojan, GMGN and pump.fun keep
Observed take rate per memecoin trading terminal, a day's fee revenue divided by that day's routed volume, per chain. What the terminal kept, not what its fee schedule says it charges.
TL;DR. As of , pump.fun posts the lowest observed take rate at 0.12% (24h avg) across 8 ranked terminals. Source: OpenChainBench, https://openchainbench.com/benchmarks/memecoin-platforms.
Read this carefully
One complete UTC day's fee revenue divided by that day's routed volume, from the public tehcscreener API, which is Allium-backed. What the terminal kept, a lower bound on what it charges: waivers and rebates pull it down. A cell reading zero is withheld rather than ranked whenever that platform charges on another chain, because this page cannot tell a waived fee from an unmeasured one. Not the all-in cost of a swap, which also includes slippage; benches 268 and 279 measure that.
Every trading terminal publishes a fee. What it actually keeps per dollar routed is a different number, and it is the one that shows up in revenue. This page divides each terminal's fee revenue on the latest complete day by the volume it routed that day, per chain. The spread is wider than any fee schedule suggests: the terminals that report fees on a chain keep between about a half and one and a tenth of a percent of what they route. Cells reading exactly zero are not ranked as the cheapest. A terminal that charges elsewhere and reports nothing here has a gap, not a price, and the page says so rather than crowning it.
Methodology
The harness reads one day per terminal per chain from the tehcscreener /bots/{bot} endpoint and publishes fees divided by volume as terminal_fee_rate_pct. Numerator and denominator are the same row of the same day, so the ratio cannot mix two periods or two chains. One column this bench used to carry is gone. Until 2026-10-07 it published a fee-paying rate, the share of a terminal's transactions that actually paid a fee, from a second metered Dune query. The current source reports fee totals and not the fee-paying share of transactions, and no arithmetic on the fields it does publish recovers it. The panel was removed rather than approximated, and the take rate it sat next to is unaffected.
Results: 8 terminals ranked by Observed take rate (24h avg)
Frequently asked
What is an observed take rate?
A day's fee revenue divided by the volume routed that day, as a percentage. It is what the terminal kept, which is not the same as what its fee page says it charges: waived fees, referral rebates and promotions all reduce the first without touching the second.
Why do some cells have no take rate at all?
Because the zero could not be explained. A terminal that charges a fee on one chain and reports exactly nothing on another, while routing tens of millions there, has a gap rather than a price, and this page cannot tell a waived fee from an unmeasured one. Rather than rank that zero as the cheapest in the market, the cell is withheld and flagged. Axiom off Solana and pump.fun on Solana are both in that state as of 2026-10-07. Volume, transactions and wallets for those cells are unaffected and still published.
Is this what a swap costs me?
No. This is the terminal's cut only. The all-in cost of a swap also includes slippage, priority fees and the pool's own fee, and a terminal with a low cut can still deliver a worse fill. Benches 268 and 279 measure the total.
Why does the same terminal show a different rate on each chain?
Because fee schedules, flow mix and competition differ by chain, and because the rate is observed rather than quoted. Pooling the chains into one number would average ratios with weights nobody chose, so each chain is read on its own.
What happened to the fee-paying rate column?
It was removed on 2026-10-07 with the source change. It measured the share of a terminal's transactions that actually paid a fee, which the current source does not report and which cannot be derived from the fields it does. It was dropped rather than approximated.
Source code github.com/ChainBench/OpenChainBench/tree/dev/harnesses/terminal-activity