How many agent-token launches still show activity after a week, a month and three months?
As of 2026-11-14, 0% of 42 virtuals tokens on base, read 30 days after first seen, showed a swap in the 24 hours before the read on a bonding pair; each launchpad and chain stands alone.
- As of2026-11-14
Virtuals launches, read from Base
One launchpad's launches, never pooled with another's.
Virtuals launches, read from Base: launches read again
Per launchpad, chain and cohort basis, the tokens of a launch cohort read again after the age shown. A cell describes what one read showed: a swap in the 24 hours before the read (on a bonding pair only where the row says so, because tokens that graduated are outside that measure) and whether the venue flags the token graduated. A cohort the lane followed as a sample says "N of M followed". A share on fewer than 20 tokens is withheld. Cohorts are added within a row (hits over tokens), never across launchpads, chains or cohort bases.
| Launchpad | Chain | Cohort basis | Age (days) | Cohorts read | Tokens | Showed a swap in the 24 hours before the read | Flagged graduated by the venue |
|---|---|---|---|---|---|---|---|
| virtuals | base | day first seen | 1 | 31 | 519 | 5.6% of 519 (bonding pair only) | 0% of 519 |
| virtuals | base | day first seen | 7 | 25 | 485 | 0.6% of 485 (bonding pair only) | 0% of 485 |
| virtuals | base | day first seen | 30 | 2 | 42 | 0% of 42 (bonding pair only) | 0% of 42 |
Our reading
Each cell describes what one read showed: whether a token of a launch cohort had a swap in the 24 hours before the read, and whether the venue flags it graduated, at the age shown.
A token with no swap in that window may have had one the day before. These shares are not a statement about any token's quality or about why it stopped. On a bonding pair, a token that graduated leaves the swap measure; the row says so.
Cohort basis matters: Virtuals cohorts are counted from the day a run first saw the launch, Clanker and pump.fun cohorts from the UTC day of the launch. The two are shown apart and never read as one thing.
What limits these figures
- Launchpads differ: what counts as a launch, a swap and graduation differs, so shares are not compared across launchpads.
- Chains differ: a launch on one chain is a different object from a launch on another.
- Sample: only part of a cohort may be followed, and the row says so (N of M followed); a share on a small sample is withheld.
- Reads are one per day. A token is read at the ages shown, not continuously.
Never pooled across catalogues, launchpads or chains; a share or price on a small sample is withheld; “unknown” means we hold no reading, never zero. Data CC-BY-4.0 — attribution: data by satohub.ai.
Reference
QuestionsWhat people ask about this pageShowHide
Questions this page answers
- What does a launch survival share measure?
- Of the tokens of a launch cohort read again at an age, the share that showed a swap in the 24 hours before that read, and the share the venue flags graduated. It describes one read, not a token's quality.
- When does the launch survival page publish?
- As of 2026-11-14, 0% of 42 virtuals tokens on base, read 30 days after first seen, showed a swap in the 24 hours before the read on a bonding pair; each launchpad and chain stands alone.
- Why are some shares withheld?
- A share on fewer than 20 tokens is not a finding, so it is withheld and the number of tokens is shown instead.
SourcesWhere the data comes fromShowHide
Sources — where launches are read
Cite this pagePlain text, APA or BibTeXShowHide
Cite this
As of 2026-11-14, 0% of 42 virtuals tokens on base, read 30 days after first seen, showed a swap in the 24 hours before the read on a bonding pair; each launchpad and chain stands alone.
Sato Hub, "How many agent-token launches still show activity after a week, a month and three months?", https://satohub.ai/numbers/launch-survival, as of 2026-11-14. Data CC-BY.
How this was measured
- Method
- Tokens launched on a launchpad are read from the chain on the day they launch and read again at set ages; each cohort cell records whether the token had a swap in the 24 hours before the read (on a bonding pair only, where the row says so) and whether the venue flags it graduated, and a share is the sum of hits over the sum of tokens followed across cohorts, withheld on fewer than twenty tokens.
- Window
- Launch cohorts read over the trailing window that holds the oldest age shown; published only once a launchpad has a month of complete days read.
- Data date
- MCP call
- none — No MCP tool returns launch survival; the cells on the page are the published form.
- URL
- none — No public export serves launch survival yet; the rows on the page are the published form.
- Recount
# one Virtuals launch record, read from the bonding contract on Base (index 0 is the first launch): curl -s -X POST https://mainnet.base.org -H 'content-type: application/json' --data '{"jsonrpc":"2.0","id":1,"method":"eth_call","params":[{"to":"0x1a540088125d00dd3990f9da45ca0859af4d3b01","data":"0x7c13774b0000000000000000000000000000000000000000000000000000000000000000"},"latest"]}'
Each cell describes one read, not a token's quality. The command reads one launch record from the contract the Virtuals lane follows; the shares follow every launch of a cohort from such records. The page is not published until the launchpads have enough history.