FOMO · Mobula Benchmark

Terminal	Median swap	Value lost, median	Lost / swap, median	App fee	Network	Pool	Protocol	Failed	p90	Swaps
·Pump Bridgeappby chainprovisional	$10.0	190 bps 1.90 %187 to 194	$0.19	66	19	93	0	—	368 bps	1752
·Binance Walletappby chainprovisional	$28.0	224 bps 2.24 %218 to 233	$0.63	50	10	135	0	1.7%	467 bps	1122
·Phantomappby chainprovisional	$15.7	247 bps 2.47 %243 to 257	$0.39	85	9	99	4	2.2%	592 bps	585
·pump.fun appby chainprovisional	$10.3	306 bps 3.06 %283 to 337	$0.31	43	5	56	122	4.6%	881 bps	803
·FOMOappby chainprovisional	$15.0	335 bps 3.35 %331 to 340	$0.50	170	21	101	4	1.0%	613 bps	4338
·BasedBotprovisional	$50.2	360 bps 3.60 %325 to 385	$1.81	59	1	156	65	2.2%	637 bps	188
·Banana Gunbotprovisional	$43.6	375 bps 3.75 %348 to 394	$1.64	99	50	61	122	3.2%	887 bps	179
·GMGNappby chainprovisional	$21.0	377 bps 3.77 %365 to 387	$0.79	99	20	68	91	1.8%	786 bps	3548
·FOMO · Mobulaapp	$42.5	—	—	—	—	—	—	0.0%	—	30

Every figure here is a median, not an average: losses have a long right tail, so averaging the sampled rows returns a larger number (pump.fun, 250 bps median against a 360 bps mean). Split columns are medians of each cost in basis points of the trade (100 bps = 1 %), each chain's median weighted by its flow on a multi-chain product; they do not sum to the median loss, the per-swap sums do (audit table).Terminal feeNetwork (tx fee + tips, origin gas)Relay (bridge fees + spread)Protocol fees (launchpad, curve, referrals)Pool (LP fee + impact, hops)

Audit the sampled swaps · the 3000 most recent transactions (0 on FOMO · Mobula), one per row with its Solscan link, loss and cost split, on the bench page

Real user swaps read on-chain (Solana: each terminal's fee-wallet and program feed; EVM: the terminals' routers and blocks read in full; 150 drawn at random per terminal per day, Solana and EVM alike, every attempt counted for the fail rate), valued at the pool's state before the trade (exact where the pool's formula is known: PumpSwap, Raydium, Launchpad, DLMM, four.meme, the EVM pools; the previous trade on the same pool within 60 s elsewhere; swaps without one keep their cost split but no loss figure). Loss = 1 − value received / value given, in basis points of the trade; terminal, network and other are exact from balance deltas, the tx fee and inclusion tips included. Published from 50 priced swaps pooled (a chain row or a single-chain product from 25), ranked from 100. Trading on Base through Mobula, the app's other bridge: the user deposits USDC on Solana, Mobula's solver buys the token on Base and delivers it. The two legs are joined on the intent id Mobula states in the clear on both of them, so the pairing is exact rather than matched by amount. Value given = the USDC credited to the solver, exact from the deposit; received = the tokens delivered at the pool's state before the settlement swap. The TERMINAL column is NOT MEASURED on this route, which is not the same as zero: no app-fee transfer exists on either leg, but a fee the app sets and the bridge takes inside its quote leaves no transfer to find, so any such charge is inside the bridge column. Pooled over the chains published on their own; a chain still filling its window is left out until then. This figure covers about 8.5 h, not the 24 h the page states, so it is published and not ranked until the window fills. Bench 268, rolling 24 h, as of Sat, 24 Oct 2026 21:05:52 UTC.