Bitquery Counts 8.98 Million Pump.fun Tokens
Roughly eight in ten tokens issued through Pump.fun, the Solana-based launchpad, never traded again after the day they went live. An on-chain study by the analytics firm Bitquery, which tracked new Solana tokens from Nov. 1, 2025 through Oct. 7, 2026, puts the share at 79% of the 8,984,141 Pump.fun tokens in its survival table. Launch day is defined as the date of a token’s first swap, and the number measures trading persistence rather than price damage: it says nothing about how far a token fell or how many holders lost money. The tail thins out fast. Only 1.5% of Pump.fun tokens registered a single trade during their fifth week, and 0.2% recorded at least 10 trades in that window. Graduation, the step at which a bonding-curve token moves to a PumpSwap pool, applied to roughly one token in 58. Supply is equally lopsided. Of about 11.5 million new Solana tokens with a confirmed first trade over the period, 78% came out of Pump.fun. Tokens that never traded at all, and a few whose first swap occurred in a long-standing pool, sit outside the sample. The methodology counts successful swaps, automated trades included, and discloses missing blocks alongside coverage gaps for newer trade formats. The figures describe trading behavior, not the share of tokens with independent holders or meaningful liquidity. Comparable launchpads land in the same range at different intensities. Launch-day-only trading reached 51% for Meteora’s Dynamic Bonding Curve tokens and 84% for Raydium LaunchLab tokens, while 1.9% of Meteora tokens and 1.3% of Raydium tokens were still trading in week five. Bitquery cautions that sample structures differ across platforms, so the gaps alone cannot establish why one venue outperforms another. That same constraint applies to the cross-platform comparisons, since each venue’s table carries its own coverage limits.
A separate September study from Jumpbit examined 1,083,771 Pump.fun launches, about 36,100 per day, using the dydt Solana trade index to follow trading through Oct. 12. It found 95.1% of launches recorded at least one trade, yet only 17.2% had a last recorded trade at least 24 hours after launch, and 7.5% at least seven days after. A mere 0.5% traded during the final 24 hours of the observation window. Jumpbit counts a creator’s own first purchase as a trade and notes that a token can keep registering activity even with its price near zero, so its numbers answer a different question than Bitquery’s survival table. Both studies count swaps, the actual exchange of one asset for another, so a token with zero post-launch swaps registers as dormant in each. Issuance shows no letup. The launch tracker MemeFees logged at least 21,765 token launches across eight platforms on Oct. 19, led by 10,570 on Flap.sh, at least 9,284 on Pump.fun and 1,707 on Pons. Its seven-day averages run 17,726 launches for Pump.fun and 11,325 for Flap.sh, with the Pump.fun counts marked as lower bounds because of incomplete collection windows. Traders are voicing the fatigue. In a post on X dated Oct. 20, an account posting as Mystral argued that “we don’t have a shortage of coins, we have a shortage of reasons to hold them,” adding that developers now expect instant volume while buyers expect instant returns. A second trader, Bishaqs, described launchpad trading as an attention market in which early buyers sell to later arrivals, and put the odds of a coin clearing the bonding curve at roughly 1 in 70. That estimate and Bitquery’s 1-in-58 graduation rate come from different samples and definitions, so neither refutes the other.
