Solana Sandwich Bots Extracted $29.7 Million Before Costs

Blockchain data provider Bitquery identified 6,850,144 sandwich attacks on Solana between Oct. 30, 2025, and Oct. 7, 2026. The attacks involved 2,545 bot wallets and affected 2,154,994 distinct wallets.

Pump.fun’s launch curve and PumpSwap accounted for 96% of the detected attacks. Bots operating outside those venues generated 39% of the estimated gross proceeds.

A sandwich attack occurs when a bot buys a token immediately before another trader’s swap and sells it immediately afterward. The bot’s purchase can raise the pool price, leaving the other trader with fewer tokens. The bot then sells at the higher price.

Bitquery calculated the estimated proceeds by valuing Solana’s SOL token at the minute of each attack. The $29.7 million figure is gross and does not account for transaction fees, failed trades or other costs. In a separate sample covering 63 attacks, bots kept 68% of aggregate gross proceeds, while 18 attacks lost money after costs. The sample does not establish net earnings for the full period.

Bitquery found a link between sandwich activity and traders’ slippage limits. Among trades routed by Jupiter through Pump.fun’s launch curve or PumpSwap, transactions allowing 20% to 50% slippage were sandwiched 14 times as often as transactions allowing less than 3% slippage.

Slippage tolerance sets the largest difference a trader will accept between the quoted price and the final execution price. A higher limit can help a trade complete during price swings, but it allows execution at a worse price before the transaction fails.

Attack rates did not rise at every higher slippage setting. Trades allowing less than 0.5% slippage were attacked more often than those allowing 1% to 3%. Trades allowing at least 50% slippage were attacked less often than those in the 20% to 50% range.

The slippage comparison covered only Jupiter routes through Pump.fun’s launch curve and PumpSwap, which represented a small share of all victim trades in the study. Inconsistent recording of Pump.fun sale fees affected the gross estimate by about 3%.

Bitquery identified bots through repeated trading patterns and excluded round trips that began within a token’s first 10 trading slots. The company reported gaps in exchange and block coverage. Blockchain records cannot establish with certainty that a wallet saw another trader’s order in advance, and some wallets in the analysis may have belonged to fast traders rather than automated sandwich bots.