Three wallets sold the old KIMCHI and bought the new one in the same minute.

01 · The vlogTwo tokens called KIMCHI

Kimchi, who posts on X as kimchi1x, put out a first vlog at about 23:00 UTC on 20 October. At the same moment, pump.fun changed its profile on X. A KIMCHI token already existed. It had launched on pump.fun in July under the name Aura Final Boss. That evening it was worth about five million dollars.

At 23:00:15 a second token appeared with the same name and the same ticker. Within the hour it was worth close to forty million dollars on paper. The old one fell from about five million to under one million in fifteen minutes. On X this is called a vamp: a fresh copy that drains the buyers of the original.

The posts that followed asked who knew. The trader zeropnl wrote that it was "physically impossible for a real human to have deployed this coin" without inside knowledge. Others pointed at the wallet that got 75% of the new supply. We asked a narrower question. Who was holding the old KIMCHI when the new one appeared, and what did they do next?

02 · The switchOut of the old, into the new

Three wallets had bought the old KIMCHI two days earlier. On 18 October they made thirty-nine buys in twelve minutes, in rounds of three. Each round was one buy from each wallet at the same instant. Together they held 5.19 million tokens.

When the new KIMCHI launched, they moved. Forty seconds after its creation they bought it, again one buy each at the same instant. Nineteen seconds later all three sold their entire stake in the old KIMCHI. The three sells landed in one block. Then they went back to buying the new one.

The switch20 October 2026, 23:00 to 23:02 UTC

They made a little on the old token and a lot on the new one. They sold the old KIMCHI into the first rush, at about five million dollars of value. Fifteen minutes later it was worth a sixth of that.

03 · Nine minutesSeven rounds in, sixty-four sells out

The buys in the new KIMCHI came in seven rounds over thirty seconds. In every round each wallet bought once, for about three SOL. By then the price was already thirty to fifty times what the deployer had paid. The wallets were not first in line. They were early enough.

The selling started less than two minutes after the last buy and was over by nine minutes past. One wallet sold in thirty pieces, one in nineteen, one in fifteen. Each came away with between thirty-three and thirty-six thousand dollars.

Seven rounds, three wallets20 October 2026, 23:00 to 23:10 UTC

Lockstep trades alone could be three copies of the same bot. The ledger says more than that.

04 · The linksThree wallets, one purse

The three wallets pay each other. Since 15 January they have made 129 direct transfers between them, in SOL and in tokens. Most go one way: 81 of them end in 3MvdJEq5vmCPemsufvg6yEGCNzFFs2Aypp4uWEY1bBW1. In twenty-six different blocks, the other two both sent to it in the same block. That is one collector and two feeders, run together.

Who pays whomdirect transfers between the three wallets, 15 January to 12 October 2026

KIMCHI is one trade among many for them. In the three days before the vlog the wallets traded about a hundred different tokens. We found no transfer in either direction between them and the new KIMCHI's deployer. We found none with the old KIMCHI's creator either, or with the wallet that got 75% of the new supply. You can check the wallets yourself with the Solana DEX API or trace them in the Bitquery MCP.

05 · BeforeThey had met this deployer before

The deployer of the new KIMCHI, EFSr396A6Jh7bNCSujGRLMkyXspbWitEvjNVatGD56oF, has launched more than thirty tokens on pump.fun since late July. Two of those launches had the same visitors.

On the last day of July the deployer created a token. Two seconds later, in one block, FHGL93 and HfN9JF both received tokens from its bonding curve. Within a minute FHGL93 began moving its share to the collector. A week later it happened again. Same two-second gap, same block for both, and both sent their tokens on to the collector in one block two minutes later.

A two-second entry is fast, and fast bots snipe many launches. What stands out is the overlap. The same wallets were early into this deployer's tokens in July and August, and in October they held the token it copied.

06 · The launchNo curve for anyone else

Most pump.fun tokens start on a bonding curve. Buyers push the price up step by step until the curve fills, and then the token moves to a pool on PumpSwap. The new KIMCHI skipped that. The deployer's launch transaction bought 793 million of the billion tokens for 85 SOL. That filled the curve in one go.

So the first price anyone else could get was 3.8 times the deployer's. Fifty-seven seconds after launch, the deployer sent 750 million of its tokens to a single wallet that has never sold. That wallet has been widely flagged on X as Kimchi's. The deployer sold most of the rest by 23:16. Its own profit on KIMCHI was about twenty-six thousand dollars.

07 · The rehearsalFour one-dollar launches first

The deployer did not start with KIMCHI that evening. At 17:54 UTC it launched a token called crime. It bought a dollar's worth and sold it fifteen seconds later. It did the same with ROCKET at 18:18, Uncle Sam at 19:44 and TRUMPINU at 21:24. Each test was bought and sold within thirty-one seconds.

Then came KIMCHI, with a first buy more than eight thousand times larger. Less than four hours after that, the deployer ran the same move twice more. A second KIMCHI at 02:41 and a token named RASMR at 02:43 each opened with an 85 SOL curve-filling buy. Both were sold off within eighteen seconds.

The deployer's evening20 to 21 October 2026 UTC

08 · Who took itWhere the first-hour money went

The ledger shows who took the first hour's money. The deployer made tens of thousands. The 75% wallet holds a paper fortune it has not touched. The cash went to the fastest traders, and the three wallets here are the clearest group among them.

They are also the group with the best seat. They held the token being vamped, and they left it within a minute of its copy appearing. We cannot see why they moved when they did. The ledger shows that they did, together, and that they have done this deployer's launches before. For other launch traces see our pieces on pump.fun Mayhem Mode and a deployer's funding trail, and the pump.fun API behind them.

09 · The recordAddresses behind this story

FAQ

What is a vamp on pump.fun?

A vamp is a new token launched with the same name and ticker as one that already trades, timed so that buyers and attention move to the copy. Here the copy launched as Kimchi's vlog went live, and the old KIMCHI lost most of its value within fifteen minutes.

Who profited from the new KIMCHI's first minutes?

Mostly fast traders. Three linked wallets bought $7,332 of it in the first minute and a half and sold it for $111,748 by 23:09 UTC. The deployer made about $26,000 from its own trades.

Did the three wallets know the new KIMCHI was coming?

The ledger cannot show what anyone knew. It shows that they held the old KIMCHI, sold all of it 59 seconds after the copy was created, and were buying the copy at the same time. Two of them also entered this deployer's launches in July and August two seconds after creation.

Are the three wallets connected to the deployer?

We found no direct transfer between them and the deployer, the old token's creator, or the wallet that received 75% of the new supply. They are connected to each other by 129 transfers since January.

How was this traced?

With Bitquery's Solana trade and transfer data. We listed the first buyers and biggest sellers of the new KIMCHI, then followed their transfers to each other and their earlier trades, including in the old KIMCHI.