Every pump.fun launch today, with the dev's record

What is in each launch

The answer is {day, complete, complete_to, count, columns, launches}; each launch is a row in the order of columns, to keep a day of 40,000 to 50,000 launches small.

Everything but migrated_at is as of the launch, counted before the coin existed, so a rule built on it sees nothing from after the launch (the point-in-time guarantee). So a rule you run on the list sees only what was known when each coin launched.

Today's launches from devs with a record

A starting filter: devs who have graduated a coin before, with 50 launches or fewer and not launching all day, on coins without a big bundle. Change the four lines to your own rule.

# Today's pump.fun launches from deployers with a record, without the serial launchers. Free, no key.
import json, urllib.request
from datetime import datetime, timezone

day = datetime.now(timezone.utc).strftime("%Y-%m-%d")
with urllib.request.urlopen(f"https://pumpdata.fun/launches/{day}.json", timeout=60) as r:
    L = json.load(r)
launches = [dict(zip(L["columns"], row)) for row in L["launches"]]
print(f"{day}: {len(launches)} launches to {L['complete_to']}")

keep = [c for c in launches
        if c["deployer_graduations_before"] >= 1       # the dev has graduated a coin before
        and c["deployer_launches_before"] <= 50        # not a launch factory
        and c["deployer_launches_last_day"] < 5        # not launching all day
        and c["bundle_supply_share"] < 0.3]            # no big bundle in the first 10 seconds
for c in sorted(keep, key=lambda c: -c["created"])[:20]:
    t = datetime.fromtimestamp(c["created"], timezone.utc).strftime("%H:%M")
    print(t, c["mint"], f'dev {c["deployer_graduations_before"]}/{c["deployer_launches_before"]} graduated,',
          f'bundle {c["bundle_supply_share"]:.0%}', "migrated" if c["migrated_at"] else "")

Today's first launches

Per launch, a wallet's first coin does better than its later ones: across the deep tables as of October 16, 2026, first launches graduated 2.67% of the time, against 0.39% for later launches by wallets that had never graduated a coin, and 1.34% to 1.68% for wallets that had. A first launch is also where the manufactured graduations hide: a quarter of graduates are bought whole by their launcher in the first seconds, usually from a fresh wallet. The bundle check screens those out. And a fresh wallet is not always a fresh person; serial operators use new wallets too.

# Today's first launches: wallets launching on pump.fun for the first time, without the buy-it-all-yourself ones.
import json, urllib.request
from datetime import datetime, timezone

day = datetime.now(timezone.utc).strftime("%Y-%m-%d")
with urllib.request.urlopen(f"https://pumpdata.fun/launches/{day}.json", timeout=60) as r:
    L = json.load(r)
launches = [dict(zip(L["columns"], row)) for row in L["launches"]]

first = [c for c in launches
         if c["deployer_launches_before"] == 0        # the wallet's first pump.fun launch
         and c["bundle_supply_share"] < 0.5]          # not bought up by its own first-10-second buyers
print(f"{len(first)} first launches without a big bundle, of {len(launches)} launches")
for c in sorted(first, key=lambda c: -c["created"])[:20]:
    t = datetime.fromtimestamp(c["created"], timezone.utc).strftime("%H:%M")
    print(t, c["mint"], f'{c["buyers_first_10s"]} buyers in 10 s, bundle {c["bundle_supply_share"]:.0%}',
          "migrated" if c["migrated_at"] else "")

For one coin's full check (its look-alikes, insiders now, top holders), open the free coin check or call GET /launch/{mint} with a free key. To test a rule over past weeks, see backtesting with /replay.

Questions

Is the pump.fun launch list free?

Yes. GET https://pumpdata.fun/launches/{YYYY-MM-DD}.json needs no key and no payment; it is limited per address like the other free pages. A whole day is about 40,000 to 50,000 launches, around 6 MB of JSON.

How fresh is today's list?

Within about an hour. Today's launches come from our live recorder, filled into the list every hour; complete_to says how far the list goes. A past day is whole and does not change.

Which fields can a backtest use?

Everything except migrated_at is as of the launch: the deployer's launches, graduations, gap since their last launch and launches in the day before were all counted before this coin existed, and the bundle is its first 10 seconds. migrated_at is when the coin reached PumpSwap, a later fact, so a rule may use it only at or after that time.

What is a first-time deployer?

A wallet with no pump.fun launch before this one: deployer_launches_before is 0. On Oct 20 and 25, 2026, one launch in six or seven came from a first-time deployer; most launches come from wallets that have launched before, many of them hundreds of times.

How far back does the list go?

The last three UTC days: today and the two before it. An older day answers 404 out_of_window, with the first day the list covers; every day since pump.fun's first is in the bulk data. A day not started yet answers 404 not_built_yet with the latest whole day.

Related: The pump.fun API · How to backtest a pump.fun strategy · The pre-buy check for a pump.fun trading agent · How many coins launch on pump.fun a day? · Free pump.fun coin check · all guides