How Many Meme Coins Are Created Every Day? 77,000 (2026)
About 77,000 a day: we counted every coin on pump.fun, four.meme, Flap.sh and Pons for a week. Most come from bots; one-coin creators graduate 7.8× as often.
77,000 meme coins a day, most of them bots. One-coin creators graduate 7.8× as often. Launch yours properly, in one transaction.
How many meme coins are created every day? About 77,000 on the four biggest launchpads, and the number is still rising. We did not take that from a headline. We counted every coin created from 20 to 26 September 2026 straight from each launchpad's contract: 246,544 on pump.fun (Solana), 131,503 on four.meme and 98,124 on Flap.sh (BNB Chain), and 65,581 on Pons (Robinhood Chain). That is 541,752 new meme coins in one week, or one every 1.1 seconds. pump.fun alone created 52,199 on 3 October.
The more useful number is who makes them. 331 wallets made 31.5% of all pump.fun coins that week, and on BNB Chain almost everything is automated. Those spray wallets almost never win. Coins from people who launched one coin that week graduated 7.8 times as often as coins from wallets launching a hundred or more. So 77,000 a day is not 77,000 competitors. Most of the field is noise, and the coins that succeed come from people with one idea and a plan for it.
This page started as a 42-minute census of pump.fun on 4 September 2026. That census is still below, with its method and its mistakes, but the headline now comes from a full week of on-chain data across three chains. The raw results and SQL are published so anyone can check them.
Disclosure: we build Alchemii, a direct-launch token creator for Solana, BNB Chain and Robinhood Chain, so we compete with bonding-curve launchpads for the same launches. Every number below is either measured from public data you can query yourself or cited to a source with its window stated.
Quick Facts
| Question | Measured answer |
|---|---|
| Meme coins created per day, four launchpads | 77,393 (20-26 Sep 2026) |
| Meme coins created per day, pump.fun | 35,221 that week; 52,199 on 3 Oct |
| Meme coins created per day, BNB Chain | 32,804 (four.meme 18,786 + Flap.sh 14,018) |
| Meme coins created per day, Robinhood Chain (Pons) | 9,369 that week; 5,948 on 2 Oct |
| pump.fun coins from wallets with 100+ launches that week | 31.5%, from just 331 wallets |
| Graduation rate, pump.fun | 3.17% within a week |
| Graduation, coins from one-coin wallets vs 100+ wallets | 9.51% vs 1.22% (7.8×) |
| Paid to pump.fun creators, September 2026 | $13,345,968, a record month |
| Paid to pump.fun creators, all time | $110.5M |
| Cost to mint your own, no code | ~0.22 SOL, ~0.32 SOL with mint revoked |
| Guaranteed creator income? | No |
How many meme coins are created every day, measured rather than estimated
Most articles answering this question cite one another, and the citations trace back to a handful of dashboards. We counted instead. Every launchpad emits a record on-chain when it creates a coin, so a full count is possible: pump.fun's create instruction on Solana, four.meme's TokenCreate event and Flap.sh's TokenCreated event on BNB Chain, and Pons V2's TokenLaunched log on Robinhood Chain. We read the first three through Dune's decoded tables and the fourth from the Pons factory ourselves.
| Launchpad (chain) | Coins, 20-26 Sep | Per day | Creator wallets | From 100+ wallets | Graduated |
|---|---|---|---|---|---|
| pump.fun (Solana) | 246,544 | 35,221 | 64,798 | 31.5% | 3.17% |
| four.meme (BNB) | 131,503 | 18,786 | 1,524 | 63.6% | n/p* |
| Flap.sh (BNB) | 98,124 | 14,018 | 5,698 | 75.0% | 0.54% |
| Pons V2 (Robinhood) | 65,581 | 9,369 | n/a | n/a | ~0.8%** |
| All four | 541,752 | 77,393 |
Scaled to everyday units, the four-launchpad count looks like this:
| Time window | New meme coins created | What that means |
|---|---|---|
| 1 second | 0.90 | A new coin roughly every 1.1 seconds, all day |
| 1 minute | 53.7 | pump.fun alone is about 24 of them |
| 1 hour | 3,225 | Four launchpads on three chains |
| 1 day | 77,393 | pump.fun 35,221, four.meme 18,786, Flap.sh 14,018, Pons 9,369 |
| 1 week | 541,752 | Every one needs an audience built from nothing |
| 1 year | 28.2 million | At this pace; it moves hard with market conditions |
How many meme coins are created on each chain?
- Solana (pump.fun): about 35,000 a day, and rising. pump.fun averaged 46,272 a day in the seven days to 3 October and hit 52,199 on 3 October itself. It is the only venue where a meaningful share of coins graduate.
- BNB Chain (four.meme + Flap.sh): about 32,800 a day. Almost all of it is automated. four.meme's 131,503 coins that week came from 1,524 wallets, and 99.3% came from wallets with ten or more launches. Our BNB meme coins report covers where the fees on that chain actually go.
- Robinhood Chain (Pons): about 9,400 a day, and falling. Pons peaked at 36,449 on 8 September and was down to 5,948 by 2 October. The Robinhood meme coins report has the daily series.
Can you trust the count?
We checked the on-chain count against an independent one. On 1 October, between 20:40 and 21:01 UTC, pump.fun's own public listing showed 1,042 new coins for our meme coin trading study. Dune's decoded create instructions for the same 21 minutes return 1,030. They agree within 1.2%. One caveat: Dune's pump.fun tables have a decoding gap from December 2025 to April 2026, so we draw no long-run comparisons across that period.
Our first method, and the 70-row trap that makes launch counts wrong
pump.fun's listing endpoint takes a sort=created_timestamp&order=DESC parameter and a limit. Ask it for 100 rows and it returns 70. The obvious pagination loop — request 100, advance the offset by 100 — therefore skips rows 70 to 99 of every single page.
That silently drops about 30% of all launches, and it is exactly the error the first version of this article shipped with. We published 17.99 launches per minute. The true rate, measured with the offset advancing by the real page size, is 24.79. Every rate on this page has been re-measured, and the script now verifies that consecutive pages abut in time and fails loudly if they leave a hole:
contiguity check: 15 pages abut exactly, no overlap, no gap
pump.fun census — 2026-09-04T21:50:29Z to 2026-09-04T22:32:45Z
tokens 1048 over 42.27 min
launches / minute 24.79
launches / day (pace) 35705
distinct creator wallets 519
median market cap $2845 (curve start = $2845)
curve complete (20min+ old) 3.09% n=517
Two more details decide whether the census numbers hold. Market caps are compared in US dollars, not in the API's quote-denominated field — 27 of the 1,048 mints are quoted in USDC rather than SOL, so that field is not one unit across the sample. And the listing runs dry past roughly 1,050 rows, which is why 42 minutes is the deepest single census available. That limit is why the headline now comes from on-chain counts: a listing shows you one evening, a contract shows you every day.
The census pace (35,705 a day, measured on a busy evening) and the on-chain average for 20-26 September (35,221 a day) land within 2% of each other, which is a good sign for both methods. Per-minute counts still swing hard: inside our 42-minute window they ran from 13 to 47.
Cross-checking the count: six sources, six different numbers
A single 42-minute window is still one window. So here is our figure against four independent counts taken at different times by people with no interest in our answer.
The spread is instructive rather than embarrassing. Marino et al. parsed 655,770 tokens created on pump.fun between 1 September and 1 October 2025 straight from Solana — 21,859 a day. The Block reported nearly 30,000 launches on 7 January 2026, a single-day record at the time. Cryptobriefing counted more than 313,000 tokens in the week to 31 August 2026 across all Solana launchpads, the highest weekly total in roughly 19 months.
Our measurements — 35,705/day from the census, 35,221/day from a full week on-chain, and 42,189/day chain-wide through Jupiter — sit at the top of that range. Read the body of evidence as a range: Solana has been minting 20,000 to 45,000 tokens a day for two years, and pump.fun has since pushed past it, to 52,199 on 3 October 2026.
How many meme coins are there in total, and how many cryptocurrencies exist?
The daily rate answers one question; “how many meme coins are there” is a different one, and the honest answer depends on what you count. The population that is tracked is tiny next to the population that was minted. On September 12, 2026 CoinGecko's global endpoint reported 21,100 active cryptocurrencies with market data across every chain. CoinGecko's own research counted 18.67 million pump.fun tokens that ever recorded a trade between January 2024 and June 2026, on one venue on one chain. Those two numbers differ by a factor of roughly 885, and the gap is the whole story: almost nothing that gets minted ever becomes something a data site bothers to list.
| Population | Count | Source and date |
|---|---|---|
| Active cryptocurrencies tracked with market data | 21,100 | CoinGecko /global, fetched 2026-09-12 |
| pump.fun tokens that ever traded | 18.67 million | CoinGecko research, Jan 2024 to Jun 2026 |
| New pump.fun tokens per day | 35,221 (52,199 on 3 Oct) | On-chain count, 20-26 Sep 2026 |
| New meme coins per day, four launchpads | 77,393 | On-chain count, 20-26 Sep 2026 |
| New Solana tokens per day, all launchpads | 42,189 | Our Jupiter sample, 2026-09-04 |
| Launchpads in DefiLlama's Solana category | 45 | DefiLlama, 2026-09-04 |
| pump.fun tokens added per year at the current pace | ~12.9 million | Derived from the 35,221/day figure |
So “how many memecoins are there” has a floor of about 18.67 million on pump.fun alone, a chain-wide figure that is meaningfully higher and uncountable without a full-chain index, and a tracked ceiling four orders of magnitude smaller. Whichever number you quote, state which population it describes.
Who is creating all these meme coins?
Here is where the obvious reading of the data is wrong, and where a second data source rescues it.
| Tokens launched by one wallet | Wallets | Share of wallets | Launches | Share of launches |
|---|---|---|---|---|
| 1 | 380 | 73.2% | 380 | 36.3% |
| 2 | 55 | 10.6% | 110 | 10.5% |
| 3-4 | 45 | 8.7% | 151 | 14.4% |
| 5-9 | 24 | 4.6% | 149 | 14.2% |
| 10 or more | 15 | 2.9% | 258 | 24.6% |
Inside the window, nearly three in four wallets appear exactly once. Read alone, that says the launch wave is a crowd of curious first-timers. It is not — that is an artefact of looking at 42 minutes. Jupiter's token feed reports each deployer's lifetime mint count, and it tells a completely different story.
The median new Solana token comes from a wallet that has already minted 48 others. Two in three come from wallets with ten or more behind them; more than four in ten from wallets past a hundred; one in seven from wallets past a thousand. The largest single deployer in our chain-wide sample had minted 173,947 tokens.
The clearest illustration sat in our own window. Wallet 3n8QxtAwHB71nd4LNme3kGML9ainK75RPtazeUJzWToK minted 43 tokens in 43 minutes, every one of them named "Glasses" — and two of the 43 finished the bonding curve. That is the strategy in its purest form: cost per attempt near zero, fire continuously, let one catch.
And it is not confined to the heavy wallets. 56.5% of every launch in the window shared its name with another launch in the same 42 minutes. Only 456 of 1,048 names were used once. "Glasses" ran 43 times, "Clippy" 35, "Working Tek" 31. Two unrelated wallets both minted a token called WOFI twenty minutes apart — GgkVueTuzbxtYKVEn7DidrbHPjL2nwtWYPG5Jfrpump at 21:52 and 95fuNysnvtu67p3gAc5erp1Enzd6SDDdxH7cwDu8pump at 22:12 — and both graduated, to $7.8M and $6.9M market caps respectively. Every mint address here is on Solscan; none of this requires taking our word for it.
So the honest description of the launch wave is not "77,000 people had an idea today." It is closer to: a few hundred ideas, cloned thousands of times, by wallets that mostly do this for a living. If you want the mechanics of that specific strategy, we documented it in how to copy a trending pump.fun coin.
Who actually graduates? One-coin creators, 7.8 times as often
The full-week data lets us test something a 42-minute window cannot: whether it matters who launches. We split every pump.fun and Flap.sh coin created 20-26 September by how many coins its wallet launched that week, then checked which ones graduated to a DEX pool by 3 October.
The answer is not close. On pump.fun, coins from wallets that launched only one coin that week graduated 9.51% of the time: 4,569 of 48,044. Wallets launching 100 or more coins graduated 1.22%. One-coin creators made only 19.5% of the coins but 58.6% of all graduations. Flap.sh shows the same pattern, even more sharply: 3.36% against 0.28%.
Our 4 September census pointed the same way (serial launchers graduated 1.97% against 2.65%) but with 1,048 coins the gap was too small to trust. With 246,544, it is the clearest result in this article.
Two honest caveats. A one-coin wallet is not always a first-timer; a serial launcher can use a fresh wallet for each coin. And causation runs both ways: a creator whose coin takes off has no reason to launch another. What the data does rule out is the spray strategy as a way to win. Firing hundreds of clones and waiting for one to catch is the worst-performing approach on both chains we measured. The coins that make it come from someone with one idea, launched once, with people waiting for it.
Do meme coin creators make money? $13.35M in one month says yes
Here is the number that makes people open a token creator in the first place, and it is real.
DefiLlama tracks what it calls supply-side revenue for pump.fun — defined in its published adapter as "creator fees paid out to coin creators and cashback returned to traders/holders of the coin." In September 2026 that figure was $13,345,968: the largest month on record, beating August's $11.31M and nearly double September 2025's $7.02M. All time, the platform has routed $110.5M to creators. On 4 September, the day of our census, $311,237 flowed out in 24 hours. For the full fee schedule and how to claim it, see how to make money on pump.fun.
The mechanism is straightforward: pump.fun charges 1.25% on every bonding-curve trade, split 0.300% to the coin's creator and 0.95% to the protocol, per its published fee page. Creating the coin costs 0 SOL. After graduation, the canonical PumpSwap pool is tiered by market cap — 1.25% total at the bottom, falling to 0.300% at the top — and the creator's slice shrinks as the coin grows, from about 1.00% down to 0.05%.
So the pool of money is genuine, growing, and at an all-time monthly high. That is the honest bull case, and the full mechanics of every income path live in how meme coin creators make money.
Now the distribution.
Why the median launch earns nothing
Averages are useless here. On census day, $311,237 of payouts divided by 35,705 launches gave $8.70 per token, and that number describes precisely zero tokens.
The median new token in our sample had a market cap of $2,845. A billion tokens priced against pump.fun's 30 virtual SOL / 1.073B virtual token initialisation comes to 27.96 SOL, which at $101.76 is $2,845 — so the median token sat at the curve's starting value, untouched, with no net buying at all. The match is exact, which is also a clean independent confirmation of the curve constants in our pump.fun mechanics explainer.
Flip the chart to its second view for the concentration. Across all 1,048 curves there was 257.7 SOL of real liquidity, about $26,220 — and five tokens held 44.7% of it, twenty held 73.2%, and the remaining 1,028 shared $7,027 between them. Creator fees are a percentage of trading volume, so that split is also the shape of who earned anything.
One finding here genuinely surprised us. 3.09% of the census cohort finished the bonding curve, counting only the 517 mints given at least twenty minutes to mature, and the full-week on-chain cohort came in at 3.17%. The most-cited academic figure, from 655,770 tokens in September 2025, is 0.63%. Ours is about five times that. Restricting to mints with thirty minutes of maturity gives 2.84%, and a different window earlier the same day gave 2.87% — the result is stable across cuts, so it is a regime change rather than noise. Graduation-rate incentives on the platform have moved substantially since the 2025 research, and any article quoting 0.63% as the current rate — including several that quote it this year — is quoting a fact about a market that has since changed.
We tested the standard advice on our own cohort. All of it failed.
Every launch guide, ours included, tells you to fill in the description, build the socials, treat the mint as a product rather than a lottery ticket. We had 1,048 tokens with known outcomes sitting in front of us, so we tested it.
| Hypothesis tested on the 1,048-token cohort | Group A | Group B | Difference | Verdict |
|---|---|---|---|---|
| Writing a description gets you more buying | 25.21% traded up (n=476, has description) | 30.24% traded up (n=572, none) | −5.03pp, wrong direction | Not significant (p=0.071) |
| Writing a description gets you to graduation | 2.73% graduated (n=476) | 2.10% graduated (n=572) | +0.63pp | Not significant (p=0.504) |
| Serial launchers do better than one-off creators | 28.99% traded up (n=407, wallet minted 5+) | 27.30% traded up (n=641) | +1.69pp | Not significant (p=0.552) |
| Serial launchers graduate more often | 1.97% graduated (n=407) | 2.65% graduated (n=641) | −0.68pp, wrong direction | Not significant (p=0.478) |
Nothing cheap worked. Tokens shipped with a written description attracted less buying than tokens without one — 25.21% versus 30.24% traded above their starting price — and while that difference is not statistically significant (p=0.071), it is certainly not evidence that descriptions help. Graduation showed a small positive difference in the other direction and it was nowhere near significance (p=0.504). Serial launchers did no better than first-timers on either measure. That last row is the one the full-week data later settled decisively, in the opposite direction from "no effect": see the section on who actually graduates, above.
We are publishing that because it is what the data said, and because the alternative — quietly running the test, getting a null, and writing the paragraph we intended to write anyway — is how most launch advice gets made.
The result is not "effort is pointless." It is narrower and more useful: the things that cost nothing do not separate winners from losers, because everyone can do them and half the field does. 45.4% of launches in our window had a description. A field where half the entrants clear a bar is a field where the bar means nothing.
What does show a difference in the published literature is the signal that costs real work. In a survival analysis of 832,941 launches, only 2.4% of launches advertised a Telegram — and those that did graduated at 1.485% versus 0.166% for those that did not, roughly a nine-fold gap. Treat it as directional rather than causal: a team that sets up a Telegram differs from one that does not in a dozen ways, and that is rather the point. The Telegram is not the cause. It is the visible residue of somebody actually having a plan, and the plan is what the market is pricing.
Where you launch decides what you keep
If you are going to be one of the 77,000 launched today, the venue decision changes your economics permanently, and it is the one decision that is fully yours before anyone else is involved.
| Where you launch | Cost to create | Cut of every trade | Who owns the liquidity | What reaches the creator |
|---|---|---|---|---|
| pump.fun bonding curve | 0 SOL | 1.25% on buys and sells | The curve, then PumpSwap | 0.300% of trade value |
| PumpSwap after graduation | 0.015 SOL migration | 1.25% falling to 0.30% as market cap rises | The protocol's canonical pool | 1.00% at the lowest tier, 0.05% at the highest |
| Direct mint + your own Raydium pool | ~0.22 SOL, ~0.32 SOL with mint revoked | 0.25% to the pool, 0% to the launch tool | You, until you burn the LP | 0.22% of trade value (0.21% on CPMM) — and 0% once you burn |
The trade-off in plain terms. A bonding curve costs nothing upfront and requires no liquidity from you — the platform handles price discovery, and in exchange takes 0.95% of every buy and sell on the curve while handing you 0.300%, then keeps a protocol slice of the graduated pool as well. A direct mint costs a flat fee once, requires you to fund your own pool, and takes nothing from your volume at any point. On a 2.5% round trip, a 0.32 SOL flat fee (itemised here) is overtaken by curve fees at roughly 13 SOL of cumulative trading volume — a threshold any coin that works at all clears in its first hour.
The row that deserves the honesty is our own. Owning the pool means Raydium routes 22 basis points of every trade to whoever holds the LP tokens — you. It also means burning those LP tokens, which is the single strongest trust signal a memecoin can produce, permanently ends that income. You cannot hold both. Bonding-curve launches never face that choice because they never had the LP in the first place; that is a genuine simplicity advantage, and it costs you the fee stream and the control.
The second difference is control, and it does not show up in a fee table. On the curve you cannot set supply, decimals, or fee tier, you cannot revoke authorities on your own schedule, and the liquidity is never yours to burn. Mint directly — through the Solana meme coin creator, or any tool that does the same job — and the authority state is settled in the same transaction that creates the token.
Is it too late to launch a meme coin in 2026?
Supply is not the constraint, and the data says why: 77,000 launches a day is not 77,000 competitors. On BNB Chain almost all of them come from a few thousand automated wallets. On pump.fun, 331 wallets made almost a third of all coins in a week and graduated 1.22% of them. More than half of the coins in our census shared a name with something minted in the same hour, and the median one never attracted a single buy. The field is enormous and mostly not trying.
But the honest version of that argument is harder than the version we would rather write, because our own null results rule out the easy answer:
| What the crowd does | Measured in our cohort | Does clearing it help? |
|---|---|---|
| Ships with no description | 54.6% of launches | No — tested, p=0.071, point estimate negative |
| Never attracts a single net buy | 72.0% of launches | This is the outcome, not the input |
| Reuses a name from the same hour | 56.5% of launches | Untested here; two WOFI clones both graduated |
| Launches from a wallet with 100+ launches that week | 31.5% of pump.fun coins | No — 1.22% graduate, against 9.51% for one-coin wallets |
| Advertises a Telegram | 2.4% (cited, 832,941 launches) | Yes — 1.485% vs 0.166% graduation |
Read that table honestly and it says one thing: the cheap differentiators are exhausted and the expensive ones are wide open. Launching one coin you actually back, with people ready for it, is the expensive one. Nobody is winning because they wrote two sentences into a form. The 97.6% of launches with no Telegram are not failing to graduate because of the Telegram; they are failing because there was never anyone on the other side of the mint. That is a problem you solve before you open the create form, and it is the whole of the marketing work.
Limitations
- Update, 2026-10-04. The headline moved from our 42-minute pump.fun census to a full week of on-chain counts across four launchpads. Launch counts for pump.fun, four.meme and Flap.sh come from Dune's decoded contract tables; Pons V2 from our own factory reads. Dune's pump.fun create tables have a decoding gap from December 2025 to April 2026, so we make no comparison across that period. 9,786 pump.fun coins have no decoded creator field; for those we used the signing wallet.
- We publish no four.meme graduation rate. The contract we read logged only 28 graduation events in two weeks, which is either a real collapse or a path we cannot see. The launch and wallet counts do not depend on it.
- "One-coin wallet" is a wallet-week measure. A serial launcher using a fresh wallet per coin would count as a one-coin creator, which if anything makes the 9.51% conservative for genuine one-off launches. It is a correlation, not proof that launching once causes graduation.
- Correction, 2026-09-04. This article first published with a rate of 17.99 launches/minute and 25,908/day. That figure came from a paginated pull that advanced the offset by 100 while the endpoint returns 70 rows per page, dropping roughly 30% of launches. Every rate, count and share on this page has been re-measured with contiguous pagination, and the script now fails if pages leave a time gap. The original figures are not preserved anywhere on this page because they were wrong, not because they were superseded.
- 42 minutes is the deepest census available, not a chosen window. The listing stops returning rows past ~1,050. Three runs across one afternoon agreed to within 1.3%, but all three sit inside a single evening on a single day.
- We measured one launchpad in depth. The chain-wide figure comes from a smaller 313-token sample through a different API. pump.fun is 87.8% of Solana launchpad fees and 90.7% of launches in that sample, not 100% of either.
- Outcomes were read minutes after minting. The graduation figure restricts to mints with at least 20 minutes of maturity, which is the honest denominator, but no 42-minute window can produce a lifetime survival rate. For that, use the cited CoinGecko and Marino work.
- The null results are null, not disproof. With n=476 versus n=572, our cohort can detect a large effect from writing a description, not a small one. A real effect of two or three percentage points would not show up here.
- Creator payouts include trader cashback. DefiLlama's supply-side figure bundles creator fees with cashback returned to traders, so $13.35M is an upper bound on what reached creators.
- Dollar figures decay. They are one SOL price from one timestamp ($101.76, 2026-09-04). The SOL amounts are the durable part.
- Out of scope: post-graduation price behaviour, buying meme coins, and whether any specific token is a rug. Those live in is pump.fun safe and memecoin FOMO.
FAQ
How many meme coins are created every day?
About 77,000 a day on the four biggest launchpads. We counted every coin created from 20 to 26 September 2026 straight from the contracts: 246,544 on pump.fun (Solana), 131,503 on four.meme and 98,124 on Flap.sh (BNB Chain), and 65,581 on Pons (Robinhood Chain). That is 541,752 in a week, about one every 1.1 seconds. pump.fun alone hit 52,199 on 3 October.
How many meme coins are there in total?
The traded population is the only countable one. CoinGecko examined every pump.fun token that ever recorded a trade between January 2024 and June 2026 and found 18.67 million, on one venue on one chain. At the pace we counted on-chain in late September 2026, pump.fun adds about 12.9 million more a year, and BNB Chain's four.meme and Flap.sh add about 12 million on top.
Who is creating all these meme coins?
Mostly bots and serial launchers. In one week on pump.fun, 331 wallets that each launched 100 or more coins made 31.5% of all 246,544 launches. On BNB Chain it is more extreme: four.meme's 131,503 weekly launches came from just 1,524 wallets, and one Flap.sh wallet made 11,478 coins. Three in four pump.fun wallets launched only once, but they made only 19.5% of the coins.
Do meme coin creators actually make money?
As a group, yes, and the amount is measurable. pump.fun paid out $13,345,968 to coin creators in September 2026, its largest month on record, and $110.5M all time, per DefiLlama's fee API. The distribution is the catch. In our census the median new token had a market cap identical to the curve's starting value, meaning zero net buying and essentially zero creator fees, while the top five tokens held 44.7% of all the SOL sitting in 1,048 curves.
How many meme coins are created on BSC every day?
About 32,800 a day on BNB Chain's two big launchpads in the week of 20-26 September 2026: 18,786 a day on four.meme and 14,018 on Flap.sh, counted from their contracts. Almost all of it is automated. four.meme's 131,503 weekly launches came from only 1,524 wallets, and 75% of Flap.sh's came from wallets that launched 100 or more coins that week.
What percentage of meme coins succeed?
On pump.fun, 3.17%: 7,803 of the 246,544 coins created 20-26 September 2026 graduated to PumpSwap within a week. That is five times the 0.63% researchers measured in September 2025. Flap.sh on BNB Chain graduated 0.54%. The rate depends heavily on who launches: coins from wallets that launched only one coin that week graduated 9.51% of the time, against 1.22% for wallets that launched 100 or more.
Is it too late to launch a meme coin in 2026?
No. 77,000 coins a day sounds crowded, but most of them are bot spam that never gets a buyer. The coins that do graduate come disproportionately from people launching one coin they mean: on pump.fun those coins graduated 9.51% of the time, 7.8 times the rate of wallets spraying 100+ coins a week. Cheap tricks such as writing a description did not help in our tests; a real audience did.
How much does it cost to create a meme coin on Solana?
About 0.22 SOL all-in through a no-code tool — roughly $22 at $101.76/SOL — covering the ~0.0191 SOL on-chain floor (network fees plus rent for the mint, token and metadata accounts) and a flat service fee. Revoking mint authority, which most memecoins do, takes it to about 0.32 SOL. Liquidity is separate: a Raydium pool needs roughly 0.4 SOL of rent plus whatever you seed, typically 5-25 SOL.
Should I launch on a bonding curve or mint directly?
A bonding curve costs nothing upfront and needs no liquidity from you, but it charges 1.25% on every buy and every sell, returns 0.300% of that to you, and never hands you the pool. A direct mint costs a flat fee once — about 0.32 SOL with mint authority revoked — takes nothing from your trading volume, and leaves the liquidity and authority settings yours from minute zero. Raydium pays 22bp of every trade to whoever holds the LP, which is you until you burn it.
The one-line version
How many meme coins are created every day has an uncomfortable answer and a useful one. The uncomfortable answer is about 77,000 a day across pump.fun, four.meme, Flap.sh and Pons, one every 1.1 seconds, most of them fired by a few thousand automated wallets. The useful answer is that the spray wallets lose: on pump.fun their coins graduate 1.22% of the time, while coins from people who launched just one graduate 9.51%. The field is enormous, mostly automated, and almost entirely undefended at the level of actually having an idea and an audience.
The money is real: $13.35M paid to pump.fun creators in September 2026, the best month yet, and $110.5M all time. It is won by the few coins with people attached. If you want yours to be one of them, the mint is the cheap five-minute part: create your meme coin →, then spend the real effort on the launch checklist and on giving people a reason to show up.
References
- Meme coin launches by chain, 30 Aug-3 Oct 2026 (raw results and SQL) — Alchemii (first-party, served from this site)Daily coin creations on pump.fun, four.meme, Flap.sh and Pons V2; the 20-26 Sep cohort split by graduation and by how many coins each wallet launched; the SQL used; and a cross-check where Dune counted 1,030 pump.fun creates in the same 21-minute window pump.fun's own listing showed 1,042. (accessed 2026-10-04)
- Decoded pump.fun, four.meme and Flap.sh contracts — DuneTables pumpdotfun_solana.pump_call_create, pump_call_create_v2, pump_call_migrate, pump_call_migrate_v2; four_meme_bnb.tokenmanager2_evt_tokencreate and tokenmanager2_evt_liquidityadded; flap_bnb.portal_evt_tokencreated and portal_evt_launchedtodex. Queried 2026-10-04. Dune's pump.fun create tables show a decoding gap from December 2025 to April 2026, so no comparison is drawn across that period. (accessed 2026-10-04)
- measure-launch-rate.mjs — the script behind the 42-minute census — Alchemii (first-party, served from this site)Paginates pump.fun's public listing contiguously and polls Jupiter's recent-token feed. Documents the 70-row page cap that made our first published rate 30% too low, and fails loudly if consecutive pages leave a time gap. No API key required — anyone can re-run it and get the same shape. (accessed 2026-09-04)
- pump.fun public coin listing (first-party measurement) (frontend-api-v3.pump.fun/coins?offset=0&limit=70&sort=created_timestamp&order=DESC&includeNsfw=true) — Alchemii, from pump.fun's public APICanonical sample: 1,048 contiguous mints created 2026-09-04 21:50:29-22:32:45 UTC. 24.79 launches/min, 519 distinct creator wallets, median market cap $2,845 (identical to the curve's starting value), 56.5% of names shared with another launch in the same window. Two further runs the same hour returned 24.51 and 24.82/min. (accessed 2026-09-04)
- Jupiter recent-token feed (first-party measurement) — Alchemii, from Jupiter's public token API313 newly created tokens across all indexed launchpads, 2026-09-04 22:25-22:36 UTC: 29.30/min chain-wide, pump.fun 90.7% of launches, Meteora DBC 3.8%. The audit.devMints field gives each deployer's lifetime mint count — median 48, with 43.5% of new tokens coming from wallets that have minted 100 or more. (accessed 2026-09-04)
- pump.fun fee and revenue summary (API) — DefiLlamaRefetched 2026-10-04 with ?dataType=dailySupplySideRevenue: creator payouts $13,345,968 in September 2026, a record, and $110,528,466 all time. Original 4 Sep pull: Fees 24h $1,309,539 · 30d $47,882,142 · all-time $1,209,069,673. With ?dataType=dailySupplySideRevenue: creator payouts 24h $311,237 · 30d $11,307,337 · all-time $96,355,495, plus the full monthly series behind the payout chart. DefiLlama defines supply-side revenue here as 'creator fees paid out to coin creators and cashback returned to traders/holders of the coin.' (accessed 2026-09-04)
- DefiLlama pump.fun fee adapter (source code) — DefiLlamaThe adapter the payout figures come from, so the definition of every bucket is readable rather than taken on trust — including the era-based protocol split (100% pre-2025-07-14, 0% from 2025-07-14, 50% from 2026-04-28). (accessed 2026-09-04)
- Solana fee overview by protocol (API) — DefiLlama45 protocols in the Launchpad category. 30d fees $54,527,428, of which pump.fun is $47,882,142 (87.8%) — close to the 90.7% launch share we measured independently through Jupiter. Creator payouts across all Solana launchpads: $12,994,927 over 30 days. (accessed 2026-09-04)
- pump.fun fees (official documentation) (pump.fun/docs/fees) — pump.funCoin creation 0 SOL. Bonding-curve trade fee 1.25% total, split 0.300% creator / 0.95% protocol / 0% LP. Graduation to PumpSwap 0.015 SOL. Canonical PumpSwap pools are market-cap tiered from 1.25% down to 0.300% total with protocol 0.050% and LP 0.200% held constant, so the creator's share falls as the coin grows. (accessed 2026-09-04)
- Protocol fees — Raydium documentation — RaydiumStandard AMM: 25bp taken on the trade, 22bp to liquidity providers and 3bp to RAY buybacks. CPMM: liquidity providers earn 84% of the swap fee, 12% goes to RAY buybacks and 4% to the treasury. This is the fee stream a creator who seeds and keeps their own pool receives — and forfeits on the day they burn the LP. (accessed 2026-09-04)
- Predicting the success of new crypto-tokens: the Pump.fun case (arXiv:2602.14860) — Marino, Naviglio, Tarantelli & Lillo — Univ. of Pisa / INFN / Scuola Normale Superiore / Univ. of Bologna655,770 tokens and 243,123 distinct creator addresses parsed directly from Solana, 1 September - 1 October 2025 — an average of 21,859 launches a day. Graduation rate 0.63% (4,338 tokens); median launch-to-graduation 4.4 minutes; 92.22% of tokens with at least 30 swaps show a statistically detected dump event. The 0.63% is the figure our own 3.09% is measured against. (accessed 2026-09-04)
- Solana trenches revive as 313K tokens created in one week — CryptobriefingPublished 2026-08-31. More than 313,000 tokens created across Solana launchpads in a single week — the highest weekly total in roughly 19 months — against $5.2B of weekly memecoin volume. States that 'roughly 1% or fewer of newly created tokens achieve any meaningful sustained liquidity or trading activity.' (accessed 2026-09-04)
- Pump.fun overhauls creator fees as token launches hit highest daily count since September — The BlockNearly 30,000 tokens launched on pump.fun on 7 January 2026, its highest single day since September 2025, alongside the creator-fee overhaul that let teams split revenue across up to ten wallets. (accessed 2026-09-04)
- Average lifespan of pump.fun tokens — CoinGecko Research18.67 million pump.fun tokens with any trading activity between 14 Jan 2024 and 18 Jun 2026. 68.67% (12.8M) recorded their last trade on the same calendar day they were created; only 4.55% survived longer than 90 days. (accessed 2026-09-04)
- Pump.fun Graduation Regime Windows: Survival Analysis of 832,941 Token Launches (arXiv:2607.02823) — KamatCited only for the social-presence effect: just 2.4% of launches advertise a Telegram, and those that do graduate at 1.485% versus 0.166% for those that do not. The paper's headline graduation rate is not used — the author's own post-publication audit found the collector observed only the first ~6 minutes of each launch. (accessed 2026-09-04)
77,000 a day, and the bots lose. Make one coin count
Coins from wallets that launched just one graduated 9.51% of the time on pump.fun, against 1.22% for the spray wallets. Mint your meme coin for about 0.22 SOL, own the pool, and put your effort into the people who will buy it.
Or launch on Robinhood Chain or BNB Chain
Related Topics
More guides to Solana token creation, authorities, liquidity, and meme coin launches.
How to Make Money on Pump.fun: Creator Rewards
How to make money on pump.fun by creating a coin: creator rewards pay 0.30% to 0.95% of every trade. See Sept 2026 payouts, Holder Rewards, and how to withdraw.
Meme Coin Market Cap 2026: $36B by Coin & Chain
The meme coin market cap is $36.45B, 76% below its record, yet launchpad fees just hit $10.3M a day. See the data by coin and chain, then launch your own.
Crypto FOMO in Memecoins: The Price of Waiting for Proof
Crypto FOMO has a price you can calculate. The median pump.fun winner graduates in 4.4 minutes — and $2,210 of buying takes 56% of the cheap supply.
Solana Memecoin Narrative Clusters: How Theme Waves Form (2026)
Seven of nine cards on one live pump.fun board were one narrative. How memecoin theme waves form, what a cluster's cap curve means, and when it's spent.
What Is Pump.fun? The Complete Platform Guide (2026)
Pump.fun Solana bonding-curve launchpad: free to create, 1.25% per trade, auto-LP at the 85-SOL (~$41K) graduation. How it works, who it's for.
How Much Does Pump.fun Make? Revenue Data (2026)
Pump.fun revenue reverse-engineered from on-chain data: daily trading fees, graduation revenue, total earnings since launch, plus methodology to verify.