Pump.fun Mayhem mode: only 1 in 50 graduations fills the pool

01 · 43 seconds on 15 JuneHow does Pump.fun's Mayhem mode work on a new coin?

Anyone can launch a new coin on Pump.fun. One of the settings at launch is called Mayhem mode. Switch it on, and a trading bot run by Pump.fun starts buying and selling your coin within seconds. Pump.fun says this should help more coins take off.

Here is what that looked like for a coin called METHEREUM. Its dev, the wallet that created it, launched it on 15 June 2026 with Mayhem on and bought half a SOL of it. A second later, Pump.fun's bot bought too. It traded the coin 23 times in 43 seconds.

Watch the amber dots. The bot's first purchase was tiny, yet the price rose 17%, when a purchase that size should have moved it by well under 1%. Its first sale knocked the price down 35%. The trades of every other wallet moved the price by the normal amount.

The dev bought more, then sold everything at a quarter of the launch price and lost about as much as the bot made. A day later the bot destroyed the tokens it had left. This dev was no beginner: it had launched 3,657 coins in the previous month.

Bitquery records Solana's DEX trades, so we followed every coin Pump.fun's bot traded from its first day in November 2025 to early October 2026, 1.75 million of them. Then we put Pump.fun's pitch to the test. Does Mayhem mode help a coin?

02 · The setupWhat is Mayhem mode on Pump.fun?

A Pump.fun coin starts life on a launch curve, which traders call a bonding curve. The curve sells the coin's tokens at a price set by a formula: every purchase pushes the price up and every sale pushes it down. Once buyers have put in about 85 SOL, it graduates. Its tokens and that SOL move into a regular trading pool on PumpSwap, Pump.fun's exchange.

Mayhem mode adds a buyer and seller of Pump.fun's own. Its explainer page says the setting mints a second billion tokens for the bot, which trades the coin for its first 24 hours "in a random walk" and then burns whatever it has left. The bot pays no trading fees. Pump.fun funds it from its own revenue and says any profit goes to an insurance fund for the bot. A legal disclaimer, rewritten in April 2026, adds limits: at most one trade a second per coin, a cap on trade size, and a balance of buys and sales set "on a discretionary basis".

Pump.fun publishes the bot's wallet, BwWK17…m2de6s. Every coin it traded counts as a Mayhem coin here. To check that this matches the setting itself, we read the Mayhem switch of thousands of random new coins straight from the chain. The bot never traded a coin with the switch off, and it traded nearly every coin with the switch on. The few it skipped fit Pump.fun's note that the bot may pass on some coins.

03 · Who uses itHow many Pump.fun coins use Mayhem mode?

For its first five months, Mayhem was a niche setting, used on under 10% of new launches. Then, in April, the share jumped to about a quarter, and it has stayed between 20% and 30% of launches every month since. Over the whole period, Mayhem coins made up 19% of the 9.0 million Pump.fun launches.

The jump came in a single day. On 8 April, Mayhem launches came to nine times the daily average of the week before, and the number of wallets launching them roughly quadrupled. Launches without Mayhem barely moved.

We found no Pump.fun announcement for that day. What we found was a fight on X. Traders posted that the bot was draining their coins, and some planned a "Mayhem Monday" to trade against it until its wallet was empty. Two weeks later, Pump.fun rewrote the disclaimer quoted above.

The new launches came from a particular kind of dev. We call a dev serial when it launched 100 or more Mayhem coins in our data. Serial devs launched 22% of them from December to February, and 71% from April to June.

04 · The main testAre Mayhem coins more likely to graduate?

Graduation is the first test a Pump.fun coin faces, so we started there. A plain count makes Mayhem look bad. Among launches from April to June, coins without Mayhem, regular coins from here on, graduated within a week 3.4 times as often as Mayhem coins.

That comparison is unfair, because devs choose Mayhem. Mayhem coins come mostly from serial devs. And they almost never have other wallets buying in their launch block, the first 0.4 seconds. That block is where snipers and the dev's own bundles show up: bots that buy the instant a coin appears, and extra wallets the dev buys through. Launches with that early buying graduate far more often, Mayhem or not.

So we compared like with like. Each Mayhem coin was set against regular coins launched in the same hour, by devs with the same launch record, with the same buying by the dev and by other wallets in the launch block. All of that happens before the bot's first trade, which comes a second or so after launch and never in the launch block. We also left out coins that graduated in their first minute, before the bot could have reached them.

Matched that way, one month stands out. Of the Mayhem coins launched in March 2026, 6% graduated within a week, 33 times the rate of similar regular coins. In every other month the gap was small, and its direction depended on how finely we matched: the Mayhem rate ran from 30% below the regular rate to 79% above it. Without March, Mayhem gave coins no clear edge.

The March graduations were not normal ones, and neither were most of the others.

05 · Hollow graduationsWhy do Mayhem graduations carry so little SOL?

Every graduation of a coin with Mayhem switched off put 85 SOL into the new PumpSwap pool. Of the 7,958 Mayhem graduations from 12 November 2025 to 5 July 2026, only 158 did, 1 in 50. All 39 in November did, before the bot began resetting prices; from December on, it was 119 of 7,919. How little the rest brought depends on when they happened. From December to February, the median Mayhem graduation brought 7 SOL. In March and April, when 72% of all Mayhem graduations took place, the median was 0.2 SOL. From May to early July it was 2 SOL.

The METHEREUM chart shows why. Each of the bot's trades also resets the price the curve quotes, nearly always by far more than the trade itself. When the bot stops, the price stays wherever its last trade left it. We read hundreds of the bot's finished curves on the chain: the median one quoted 28% of a fresh coin's price, and most quoted less than half. Nearly all regular curves, 97% in our sample, sit within 1% of the launch price.

At a low price, buying every token left on the curve costs a few SOL instead of 85 SOL. That last purchase sets off a graduation, and the new pool opens with only the SOL buyers paid in. This took off in March, right after the bot changed the way it traded on 10 and 11 March.

Graduations after early July, past the end of Pump.fun's graduation records in our data, look the same. Of 31 Mayhem graduations from July to September that we checked on the chain, 26 put less than one SOL into the new pool.

Counting only graduations that filled a normal pool, the matched comparison is one-sided. Similar regular coins were at least 22 times as likely to reach a full pool in every month. Over the whole period the gap was 48 to 91 times, depending on how finely we match. The result holds when we match on the launch day instead of the hour, on the hour alone, or add the buyers of the first minute. It holds for the same devs, too: in the 44,051 weeks when a dev launched both kinds of coin, the dev's own regular coins filled a full pool 67 times as often.

06 · Real buyersDo buyers show up for Mayhem coins?

Pump.fun's pitch is that the bot's trading draws attention, and attention draws buyers. On the same matched launches, it did not. Wallets other than the dev and the bot bought more SOL of similar regular coins on their first day than of the bot's, in every month and under every matching we tried. How much more depends on the month and the method: from 1.2 to 14 times as much, and four times as much over the period with our full match. The regular ones were far more likely to draw a crowd of 100 or more outside buyers. The share still trading on day two was about the same.

Some of those outside buyers are bots themselves. Sandwich bots, for one, buy just before the Mayhem bot's orders and sell just after them, and they bought 6% of the outside SOL on the bot's coins. Taking them out would widen the gap.

07 · The botHow long does the Mayhem bot trade a coin?

Pump.fun says the bot trades a coin for its first 24 hours. In practice it was gone within a minute on 61% of them. From its first trade to its last, the bot spent a median 39 seconds on a coin, and was done within a few minutes on nine in ten. It never stayed for a full day.

Until March, the bot stopped at exactly 50 trades on about half of all coins. That cap went away on 10 and 11 March. Pump.fun's disclaimer promises no more than one trade a second on a coin. On 15 September, 31% of the bot's repeat trades on a coin landed in the same second as its previous one.

Short as each visit is, the bot visits a lot of coins. It made 92.8 million trades over the period, and by September it made 18% of all trades on Pump.fun's launch curves. Its trades are small, so in September it made only 5% of the SOL traded there.

08 · The costDoes the Mayhem bot make or lose money?

In April, traders on X accused the bot of taking more than $1 million from them in a single day. Over the full period, the bot lost money. It put more SOL into launch curves than it took out. It won some back on PumpSwap, where it also trades coins that have graduated. On top of that, it paid network fees and tips to Jito, a service that gets transactions into blocks fast.

All in, the bot is down about 39k SOL, roughly $3.5 million at the SOL price of each day. Pump.fun pays that from its own funds. As for the extra tokens minted for it, the bot sold about 5% of them for good and burned the rest. Once the bot is done, most coins end up close to the normal supply.

09 · Who gainsWho ends up with the Mayhem bot's SOL?

When the bot puts in more SOL than it takes out, that SOL goes into the coin's curve, and whoever sells into the curve takes it home. We followed every other wallet's trades on the bot's coins launched up to the end of June, over each coin's first week, and counted what they paid and got after Pump.fun's trading fees. Devs also earn a creator fee of 0.30% on most coins' trades, the bot's excepted, so we added that to their side.

No group of traders came out clearly ahead. The 55,610 devs who tried Mayhem once lost about 13,500 SOL between them, and devs with 2 to 99 coins lost too. The 1,372 devs who launched 100 or more, 552,557 coins in all, roughly broke even: about 3,800 SOL ahead, under 0.01 SOL a coin. Depending on how many of their coins pay them a creator fee, the true figure sits between a loss of 500 SOL and a gain of 4,500 SOL. A little over half of these devs (58%) ended ahead. The biggest single winner, 3LUfv2…RR2Yww, launched 2,287 of these coins and kept about 2,200 SOL.

Everyone else put about 61,600 SOL more into these coins than they took out, after fees. So the bot's SOL did not end up with traders as a group. It stayed in the coins, in curves that never filled and in the thin pools of those that graduated, and some of it went to Pump.fun as fees. What sandwich bots take around each of the bot's orders is counted in our Solana sandwich attack study.

10 · TimelineWhat changed, and when?

The bot did not behave the same way across the 11 months. The dates below come from its trades and from the coins it left behind.

11 · For buyersWhat should you check before buying a Mayhem coin?

If a coin has Mayhem mode on, then for its first minute or so a Pump.fun bot moved the coin's price, and moved it by far more than its own trades. A sharp rise in that window may be the bot. When the bot is done, the coin's curve often sits below its launch price.

A graduated Mayhem coin is not the same as a graduated regular coin. Before buying one on PumpSwap, check how much SOL its pool holds. Bitquery's Pump.fun API returns every trade on a coin from launch, including the bot's, and the PumpSwap API shows the pool's trades after graduation. For the wider picture of how Pump.fun coins fare, see our Solana launchpad study.

12 · MethodHow did we measure this?

We used Bitquery's records of Solana DEX trades from 30 October 2025 to 7 October 2026, Pump.fun's launch and graduation records up to 5 July 2026, and Solana transfers up to 30 August 2026. The full history of Solana trades is also sold as files on the Bitquery Data Store, and the trades are available through Bitquery's Solana DEX API.

Mayhem coins. A Mayhem coin is a Pump.fun coin that Pump.fun's published bot wallet traded on its launch curve. Pump.fun stores the switch on the chain, in each coin's curve account. We read it for 3,600 random coins launched from November to early July and 1,600 launched later. The bot traded no coin with the switch off. In the first sample, it traded 565 of the 585 coins that had it on.

Graduations. A coin graduates when Pump.fun moves it to a PumpSwap pool. We count a graduation only when the move and the creation of the pool happen in one transaction, and we read how much SOL the move put into the pool. Pump.fun's graduation records in our data end on 5 July, so the matched comparison uses launches up to 27 June and a 7-day window.

The matched comparison. We took launches priced in SOL from 1 December 2025 to 27 June 2026. Each was compared with regular coins launched in the same hour whose dev had the same record: launches in the previous 30 days (five bands) and whether any had graduated. The dev's purchase in the launch block had to match too (six bands), and so did the buying by other wallets in that block (five bands). We report the average, over Mayhem coins, of the gap to their own group. 92% of them had a group. We left out launches from 7 to 26 May, when part of Pump.fun's trades is missing from Bitquery's records, and coins that graduated in their first minute, before the bot could reach them.

Outside buyers are wallets other than the coin's dev (the wallet that created it, or that paid for the dev's trades) and the bot.

The bot's money is the SOL in and out of its wallet on each trade, with coins priced in USDC converted at each day's SOL price, plus the network fee of each of its transactions. Jito tips were measured on the 15th of each month and scaled to the month's trades, so that line is an estimate.

Who gains. For the bot's coins launched up to 30 June, we added up every other wallet's trades on the launch curve over the coin's first week. Sandwich bots, snipers, dev bundle wallets and other bots come from the wallet lists built for our studies of snipers and bundles, sandwich attacks and real Solana traders.

Checks. On a public Solana node we decoded the bot's trades, which showed no fee and the price reset after each trade. We also decoded 6 graduations of its coins, where the SOL in the new pool matched our records to six decimals, and 972 finished curves.

FAQ

What is Pump.fun Mayhem mode?

Mayhem mode is a setting a dev can switch on when launching a Pump.fun coin. Pump.fun then mints a second billion tokens for its own trading bot, which buys and sells the coin during its first day and burns whatever tokens it has left at the end. The bot pays no trading fees.

Is Mayhem mode good for a Pump.fun coin?

Not by the measures we checked. Apart from a spike in March 2026, Mayhem coins graduated about as often as similar regular coins, and 98% of Mayhem graduations put far less than the normal 85 SOL into the new pool. Counting only full graduations, similar regular coins did at least 22 times better in every month.

How long does Pump.fun's Mayhem bot stay on each coin?

Pump.fun says up to 24 hours. In practice the bot's first and last trade on a coin were a median 39 seconds apart, and on nine in ten coins it was done within 4 minutes.

How much does Pump.fun's Mayhem bot make?

It loses money. Over our 11 months it lost about 39k SOL after network fees and Jito tips, roughly $3.5 million. It lost most on the launch curve and won some back on PumpSwap.

What does it mean when a Pump.fun coin graduates?

A Pump.fun coin graduates when buyers have taken every token its launch curve sells. For a regular coin that takes about 85 SOL, and the tokens and SOL then move to a PumpSwap pool. A Mayhem coin can graduate with far less, because the bot leaves its curve at a lower price.

How can you tell if a Pump.fun coin has Mayhem mode?

On the chain, the setting is stored in the coin's curve account, and the coin's early trades include many from Pump.fun's bot wallet, BwWK17cbHxwWBKZkUYvzxLcNQ1YVyaFezduWbtm2de6s.

Why did Mayhem launches jump in April 2026?

On 8 April the number of wallets launching them went from a few hundred per day to over 1,600, and most of the new coins came from devs who launch by the hundred. Counting fees, those devs roughly broke even on their Mayhem coins. Pump.fun made no announcement that we could find.