Read the create block
The block of the create transaction and the two after it, decoded from pump.fun's own trade events: every buy, every failed attempt, in block order.
Who bought before you could?
Every launch opened at slot 0. See which buyers were linked to the creator, which weren't, and what every dev did before. Every link proven on chain.
Today· 5 launches· 20.0% clean· 60.0% linked slot-0 buys· updated 3002607s ago
Live board
Slot —Launches today
5
+5 last hour
Clean rate
20.0%
7-day average after the first full day
Supply bought by linked wallets
12.7%
avg in slot 0 per launch with linked buys · 3 launches
Launches per hour
| period | launches | clean | linked buys | sniped |
|---|---|---|---|---|
| 2026-09-21 14:00–15:00 UTC | 5 | 1 | 3 | 2 |
cleanlinked buyssnipedavg clean
UTC hours.
Inspect a launch
Mint…1111 · ago
Vector ELINKED BUYSSNIPED
Creator0.5%
Linked0%
Cluster9.0%
Unlinked12.5%
How it works
Three steps, every launch
Trace who funded every buyer
SOL sent between each buyer and the creator, shared funders, and tokens passed back within 24 hours. Exchange hot wallets are never a link.
Label the launch and the dev, with proofs
Clean, linked buys, sniped, dev-heavy or quiet, each with its reason and the transactions behind it, then the dev's record across every launch.
Live feed
New launches, as they are read
One at a time, so it stays readable. Market cap is read one hour after launch.
| time | token | labels | slot 0 | linked | unlinked | dev initial | proofs |
|---|---|---|---|---|---|---|---|
| Vector E VECE | LINKED BUYSSNIPED |
9.0% | 12.5% | 0.5% | proofs → | ||
| Vector D VECD | LINKED BUYS |
15.0% | 0% | 3.0% | proofs → | ||
| Vector C VECC | LINKED BUYS |
14.2% | 1.0% | 2.0% | proofs → | ||
| Vector B VECB | CLEANSNIPED |
0% | 11.3% | 1.5% | proofs → | ||
| Vector A VECA | DEV-HEAVYQUIET |
0% | 0% | 8.0% | proofs → |
Cleanest devs this week
All devs →At least 3 launches in 7 days, every one CLEAN, ranked by launches still trading after 24 h.
No dev in the test vectors qualifies.
Biggest linked buys today
All →Share of supply bought in slot 0 by wallets linked to the creator or funded together.
The rules we use
Methodology →- CreatorThe creator wallet; its buy inside the create transaction is the dev initial buy.
- LinkedOne concrete on-chain link to the creator, shown with the transaction that proves it.
- ClusterNot linked, but three or more slot-0 buyers funded by the same non-exchange wallet.
- UnlinkedEveryone else in slot 0, typically snipers.
Builders
Query any launch from your agent
A remote MCP server with four read-only tools: check a launch, a dev's record, the day's numbers, recent launches. Or the REST API, 60 requests a minute, no key.
API & MCP docs →claude mcp add --transport http block-zero https://blockzero.site/mcp
Questions
What is slot 0?
The slot (about 400 ms of Solana) that holds the transaction creating the coin. Buys in the same slot happen before almost anyone could see the coin exist. Block Zero reads that block, and the two after it, for every pump.fun launch.
What does “linked” mean?
A concrete on-chain link between a buyer and the creator, shown with the transaction that proves it: SOL sent between them in the week before launch, funding from the same wallet that funded the creator (exchange wallets excluded), or tokens sent to the creator or a linked wallet within 24 hours.
Can a record be wrong?
Yes, within stated limits. Funding older than 7 days, trails through exchanges, and very busy wallets beyond the read cap are not seen; a partial record says what could not be read. The method and every limit are on the methodology page.
How do I dispute a record?
Every launch and dev page has a “Dispute this record” link. A person reads every dispute; a record found wrong is corrected or annotated, with the reason logged.
Is this advice?
No. It is a record of what happened on chain: who bought in the first slot, how they were funded, and what they did in the first day.
Which launches are covered?
Every pump.fun create since the indexer started, found from chain data (a WebSocket plus a gap-free reconciler), and older launches of any dev someone looks up. Measured coverage is published on the methodology page.