Chainlytics Verify checks what trading screens say about a new token against Solana mainnet, and tells you who is on the other side of the trade: the creator, the wallets that bought with it, and whether it is one of the biggest token factories.
Live check demo against Solana mainnet. Full launch rebuild for pump.fun launches; tokens from other launchpads get the authority, supply and source checks for now. Up to a minute per token, limited to a few checks per minute.
Live on Solana. Next: Base, BNB Chain, Ethereum and Robinhood Chain.
The launchpad market in four numbers, from daily public data on every major launchpad and trading terminal, checked against the chain where we can.
Market figures: tehcscreener.com public API (data by Allium and Dune), analysed by Chainlytics Verify.
Most of what decides a launch happens before a human can read the chart, and the wallets doing it change every week. Reading it at this speed and scale takes a model trained on how launches are built.
The creator mints and often buys in the same transaction.
Wallets that land in the creation block. Who funded them?
Bots that buy before any chart exists.
On a large share of launches the creator has already sold.
Screens, calls and trending lists bring buyers in.
Is there an informed way to play this market? That is what we test, one pre-registered hypothesis at a time, on every launch. Each hypothesis is written down before the data is read and read once; subscribers get every result. Nothing here is a promise of profit.
Pump.fun turned launching a token into a one-click act: no code, no liquidity to seed, a bonding curve that prices every trade from the first second. It became one of the highest-earning products in crypto, passing $1 billion in revenue in about two years, and tens of thousands of tokens launch on it every day.
That revenue is paid by the people who trade those tokens. Most of it goes to buying back the platform's own token, and creators take a cut of every trade. What reaches the buyer is the part nobody audits. Public data points one way: Bitquery attributes at least $579 million of profit to developer-linked and bundled launch wallets, against $31.5 million for sniper bots. Our own pre-registered tests, below, point the same way: launches built by the same playbook cost buyers more, and the wallets behind them change while the playbook does not.
That is why Pump.fun is a case to audit, not a casino to guess at. Chainlytics Verify reads every launch on-chain from its creation transaction and shows who is on the other side (the creator, the wallets that bought in the same block, where the money came from and where it goes) before you become the exit liquidity. We don't tell anyone what to buy and we promise no profit. We measure where the money goes.
Revenue: DefiLlama via The Block, March 2026. Launch-wallet profit: Bitquery, October 2026 (launches to June 30, 2026; a lower bound).
We fixed one simple trading rule before computing any return, including realistic round-trip costs, and applied it to both groups. Every top deployer active that week had a negative mean, from -21.4% to -7.6%.
Wallets are cheap to replace; the way a launch is built is not. We trained a model on how the top deployers build their launches, with no wallet identity at all, then applied it to other wallets in weeks it had never seen.
The pass rule was written down before the data was read, and the test ran once, on about 3,400 launches from other wallets. A second, separate week gave the same gap.
Flagged launches almost never went quiet. They draw buyers in, and those buyers lose more.
Screens and the chain often disagree under the same label: holder count, top-10 share, the dev wallet and the price on the curve. Verify shows the chain value and the definition behind each number.
In 18% of random launches the dev and the launch buyers took 10% or more of supply, and a popular risk scanner rated 82% of those lowest risk.
Median 1.05% of supply.
Of what it bought at launch.
Median 6.6% of supply.
Every link comes with the transactions behind it.
We counted every pump.fun launch on-chain, day by day for four weeks, and compared it with a widely used public dataset. Most days matched almost exactly. Some did not, and nothing on the dataset said so.
Within 0.05% of the on-chain count.
About one launch in five missing from the public count.
Our own stream had one short gap too. Cross-checking shows exactly where each source fails.
Give Verify a mint. It reads Solana mainnet, compares what the screens report and rebuilds the launch. Before you buy, it counts the red flags on the launch and shows what happened to outside buyers on similar launches, as history, not a prediction. Every answer is stored with the raw data it used, so anyone can check it.
Scanners give you an opinion. Verify shows you the chain value, the definition behind each number, and a receipt you can replay.
A list of wallets goes stale the day a deployer moves to a new one. More than keeping a database of wallets, our work goes into making the model trained on launch craft more robust every day, so it keeps catching new malicious launch strategies, whoever runs them. That is the technology moat of Chainlytics Verify.
A model that learns how malicious launches are built, not who builds them. It found other wallets' launches built the same way, in weeks it had never seen (Finding 2).
The scoring design behind the whole platform: nine factors that act as gates across three distinct phases of a launch.
Launchpads now pay users to "call" tokens to their followers. The payouts are public. What the followers earn is not, so we are measuring it.
Every public callout as it is posted, with the caller's wallet. For each call we check the chain: did the caller hold the token before calling, did the caller sell within 30 minutes after the call, and what a follower who bought one minute later had one hour later, after costs.
The rules were written before the data, and the result is read once. Percentages, no names and no accusations: every figure is backed by transactions we keep. Pro members get the full result first; the headline goes on this page.
Payout totals: crypto.news, citing the co-founder and Sparsity Research. Callout data: public callout feeds.
No signals, no calls, no promise of profit. What happened on the launchpads, what it cost the people trading, and which public numbers you can trust, checked against the chain.
Founding members: the price is locked for 12 months, and billing starts only when your first report is delivered. Cancel any time.
Founding member places for Pro and Data are open. Send us a DM on X.