Reports

Five of every six token creations are not launches

Published 15 September 2026. The figures below were computed from the record on that date and are not updated afterwards: they are read from a file written when this report was published, not recomputed when this page is built. The live view is here.

Since 2026-09-11 this archive has read Solana's blocks directly - not a launchpad's feed, the chain itself - and recorded every token creation it found. By 22:27 UTC on 2026-09-15 that was 1,020,962 mints, one about every 0.33 seconds, averaging 257,921 a day across the window. Of those, 170,694 - 16.7% - are launches: a token created by an identified launch program to be sold as itself. The other five in six are the chain's plumbing wearing the same data type: pool tokens, positions, receipts and wrappers.

These figures describe the SCANNER'S WINDOW, four days of it, not the archive. The launch record itself extends back to 2026-09-02 and holds far more launches than this window contains; the front page counts that record. This report is about what four days of reading every block shows: how token creation divides between launches and everything else.

The largest single creator of tokens in the window produced 756,294 mints - 74% of everything created - and not one launch. A count of token creations is therefore not a count of anything a person was asked to buy, and any figure that treats the two as one number is measuring infrastructure. Separating them takes an identification per program, with evidence, which is what the launch programs page exists to hold.

The labelled programs are identified on the launch programs page, each with the evidence for its identification. An unlabelled program is one we have not identified, and a zero in the launches column is a statement about what that program created in this window, not about what it is. Of the 170,694 launches, 160,495 carry the name they declared and 159,252 (93.3%) have their launch document captured and hashed.

The window runs from slot 446,273,941 to 447,355,810. The scanner read 1,081,276 of the 1,081,870 slots in it - 99.945% - and the missing 594 slots are recorded as gaps, published live as ranges on the status endpoint, never filled in by interpolation. That coverage was not free: on 2026-09-14 a restart fault cost a sixteen-minute gap, and on 2026-09-15 Solana began carrying version-1 transactions and the scanner silently fell to a twentieth of its throughput until the cap was found and lifted the same night - after which a backfill re-read every recoverable slot while the public nodes still served them. 273 of the missing slots are from 2026-09-11 and are now beyond public RPC history: a permanent, known, bounded hole.

It does not grade any program. A program whose mints are not launches is infrastructure doing its job, and a zero in that column accuses nobody of anything. It does not claim the launch identifications are complete: the remainder includes 23,705 launches from programs not yet individually identified, and identification only ever adds. And every count here is a floor - a mint created inside one of the recorded gaps is absent from all of them.

The figures above were read from the chain scanner's own database, which is not a published file: the launch layer it identifies is what flows into the record, and the coverage it claims is published live as slot ranges at /api/v1/status. The chain-wide figures reproduce from the chain itself - read every block between slots 446,273,941 and 447,355,810 from any full node and count the mint initializations - which is the one form of checking that depends on nothing we serve.

Since 13 September 2026 the record database is licensed and downloads with an API key. Text above written before that date is left as published.