Dead and dying memecoin launchpads: Heaven, Believe, Noxa and what killed them

Heaven, Noxa and LetsBonk are finished as going concerns; Believe's operating status cannot be confirmed. Each died differently — a reflexive buyback flywheel unwinding, an abrupt disappearance, a slow bleed after losing share, a payments failure. Vector is often listed alongside them and should not be: Coinbase acquired it.

Nobody writes obituaries for launchpads, which is why the wrong ones keep appearing in "best launchpad" listicles two years after they stopped processing volume. This page fixes that as of 31 August 2026.

How to tell a dead launchpad from a quiet one

The vital sign is protocol fees, not token price, social activity or a working website. A launchpad that is live but unused still renders a homepage and still has a Telegram. It does not collect fees.

DefiLlama publishes 24-hour, seven-day and 30-day fee figures per protocol, plus a quarterly trend. Three readings matter. Fees at or near zero over 30 days means nothing is being traded on the venue's own contracts. A missing protocol page entirely — a 404 — is a stronger signal still, because it usually means the adapter was removed for lack of activity. And a quarterly series falling by an order of magnitude each quarter tells you the trajectory rather than the snapshot, which matters because any single day can be an indexing artefact. Treat 24-hour numbers as soft and prefer seven-day and 30-day figures.

One more distinction worth holding: a launchpad can be dead for issuance while its already-launched tokens keep trading elsewhere. Those are separate facts and conflating them produces both false obituaries and false survivals.

Heaven: the God Flywheel unwound

Heaven is the cleanest case study in the category because it died of its own design rather than of neglect.

As of 31 August 2026 its fees were $0 across 24 hours, seven days and 30 days. DEX volume over the same windows was $177, $293 and $10,541 against cumulative volume of $412.28M. The quarterly fee series is the whole story: $4.48M in Q3 2025, $120K in Q4 2025, $24.8K in Q1 2026, $7.3K in Q2 2026, and $0 in Q3 2026. That is not a decline, it is a collapse with three quarters of aftermath.

Mechanically, Heaven was unusual for 2025. It had no bonding curve. Each pool was seeded with roughly 35 SOL of "virtual liquidity", and trading fees ran at 1% below a $100K market cap, then split into 0.25% to the community and 0.5% to the creator above it. The distinguishing feature was the "God Flywheel": 100% of protocol revenue was directed into buying and burning LIGHT, Heaven's own token.

The flywheel worked exactly as designed, in both directions. LIGHT went from roughly $15M market cap to about $130M, then back to around $33M. The logic is worth stating plainly because it recurs constantly. When launch volume rises, revenue rises, buy pressure on the platform token rises, the platform token's price rises, which attracts more creators, which raises launch volume. Every term in that loop is also a term in the reverse loop. Nothing in the mechanism distinguishes between organic demand and demand caused by the buyback itself, so the structure amplifies whatever direction it is already moving in.

Reflexive buyback flywheels are not fraud, and Heaven's did what its documentation said it would. They are simply an amplifier attached to a variable nobody controls. Read any launchpad that routes a large fraction of revenue into its own token accordingly.

Noxa: gone in eleven days

Noxa.fun was Robinhood Chain's first significant launchpad and its shortest-lived. It halted new launches on 11 July 2026 and the site went dark on 13 July 2026, having taken roughly $12M in fees across about 60,000 launches in under two weeks — roughly 75% of all token deployments on the chain during that window.

The exit sequence was disorderly. Noxa first blamed bot spam, then announced it was routing 100% of trading fees to creators, then cited a "Cloudflare issue" as the site went down. The community split between calling it FUD and calling it a soft rug, and no conclusive finding of fraud has been established. CASHCAT, the chain's largest token, fell 33% in 24 hours on the exit. Residual trading of around $10.5M per 24 hours continued on tokens Noxa had already launched, but the venue issues nothing new. Days later a rival, Vlad.fun, also went offline citing an "internal integrity issue."

Noxa should never be described as a live launchpad. The full sequence, and what it revealed about how quickly launchpad share reallocates, is in the Noxa collapse explained.

Believe: status unconfirmed, and that is the honest answer

Believe was one of 2025's higher-profile consumer launch products and it does not have a DefiLlama protocol page — the URL returns a 404. For a protocol that once had meaningful volume, a missing page is diagnostic rather than neutral.

What is documented is a failure path rather than a shutdown. Believe paused on-chain payouts and asked creators to move to PayPal, and founder Ben Pasternak faced litigation over the token's collapse. Both of those are verified.

What is not documented is whether Believe is currently operating. We could not confirm it either way as of 31 August 2026, and we are not going to assert a shutdown that no source states. If you are a creator with outstanding payouts, that ambiguity is the operative fact: the payout rail changed, the on-chain guarantee went away, and you are now an unsecured creditor of a company rather than a holder of an on-chain claim. That distinction is the entire reason on-chain payouts were the selling point in the first place.

LetsBonk: technically live, commercially dead

LetsBonk (bonk.fun) is the largest reversal in the sector's short history. In July 2025 it held roughly 54% of Solana launchpad share. Its own official revenue dashboard in August 2026 showed monthly revenue of $84,910, a daily average of $2,740, 24-hour volume of $5,410 and 24-hour fees of $380.76, against a treasury of 810 SOL and 16,295 USD1.

The tell arrived before the numbers did. On 3 December 2025 LetsBonk restructured its revenue split, removing the Buy/Burn and SBR mechanisms entirely and replacing them with a "Buy for BNKK" allocation at 51%. A launchpad removing its own flagship burn mechanism is a statement about expected future revenue, made by the people with the best information. It is technically live and commercially finished. The full arc is covered in the rise and collapse of bonk.fun.

Zora: killed by its own distribution partner

Zora is a slightly different animal — creator coins rather than pure memecoins — but it belongs here because the shutdown signal was unusually explicit. Brian Armstrong, on 13 July 2026: "It didn't work. We pivoted earlier this year. We messed up, time to move on." Base had already discontinued Creator Rewards and removed the social feed in February 2026.

The numbers follow the statement. ZORA fell roughly 95%, from about $550M market cap in August 2025 to around $30M. DefiLlama on 31 August 2026 showed 30-day fees of $13,775, 30-day revenue of $5,660, 30-day DEX volume of $504K and TVL of $5.55M. At peak, 1.6M creator coins had been minted against $470M of volume by August 2025. The post-mortem is in Base's creator-coin experiment, ended.

What killed Zora was not a mechanism failure like Heaven's or a disappearance like Noxa's. It was distribution withdrawal. The platform depended on Base surfacing its content, and when Base stopped, there was no independent demand underneath.

Vector was acquired, not shut down

This correction matters because the two look identical from outside: the apps wound down and the product stopped existing.

Coinbase acquired Vector.fun from Tensor Labs, announced 21 November 2025, price undisclosed, its ninth acquisition of 2025. The apps were wound down and the Solana-native technology was folded into Coinbase's DEX trading integration. Founders Richard Wu and Ilja Moisejevs joined Coinbase along with 11 employees. Tensor Foundation retained Tensor Marketplace and TNSR independently.

An acquisition and a collapse produce the same dead URL and opposite signals about the team, the technology and anyone who dealt with them. Any list that files Vector under failures is not reading the primary sources.

The residual category: no data, not proof of death

Boop, Time.fun and Daos.fun still appear in listicles. None of them has 2026 data of substance, and none appears in DefiLlama's launchpad fee rankings. Daos.fun in particular shows up repeatedly with no volume attached to it.

The honest framing is that these are residual rather than confirmed dead. Absence from a fee table is weak evidence — indexing coverage is incomplete, particularly for smaller and Telegram-native products — but combined with an absence of 2026 coverage it is enough to say you should not build anything on them without checking activity yourself first.

What this doesn't tell you

Fee data measures a venue's own contracts. It does not measure whether a team is still employed, whether a treasury is solvent, whether creator payouts are outstanding, or whether tokens launched on a dead venue still trade — several do, in volume.

DefiLlama's coverage is also uneven. It indexes AMM and launchpad contracts well and Telegram-native products poorly, so an absence from its tables is an inference, not a finding. Where a protocol page is missing entirely, as with Believe, we have said so rather than converting a 404 into a conclusion.

And nothing here predicts which currently-live venue is next. Heaven's quarterly series was already falling hard two quarters before it hit zero, which is a pattern worth watching, but Noxa went from roughly 75% of a chain's deployments to dark in eleven days with no such warning. Per-venue launch, graduation and fee data is published on the Meme Central analytics hub so you can check the trend yourself rather than trusting a listicle's publication date.

What actually protects you

Very little of this is preventable at the buyer level, but two habits reduce exposure. Check the venue's fee trend before you launch on it or buy something new from it — a launchpad with collapsing revenue has diminishing reason to maintain infrastructure you depend on. And prefer arrangements where the guarantee is on-chain rather than contractual, which is precisely the distinction Believe's creators discovered when payouts moved to PayPal. A locked liquidity position survives a platform going dark; a promise from the platform does not. What that lock does and does not cover is set out in what a liquidity lock does not protect you from.

Frequently asked questions

What happened to the Heaven launchpad?

Heaven's revenue collapsed alongside LIGHT, the token its "God Flywheel" bought and burned with 100% of protocol revenue. Quarterly fees fell from $4.48M in Q3 2025 to $0 in Q3 2026, with $177 of DEX volume in the 24 hours to 31 August 2026. LIGHT ran from roughly $15M to $130M market cap and back to about $33M.

Is the Believe app dead?

Unconfirmed. Believe has no DefiLlama protocol page, which for a formerly active protocol is diagnostic, and it verifiably paused on-chain payouts and moved creators to PayPal while founder Ben Pasternak faced litigation. But no source we could verify states that it has shut down, so we do not assert it. Do not rely on outstanding payouts.

Is LetsBonk still working?

Technically yes, commercially no. Its own dashboard showed $84,910 monthly revenue and $5,410 of 24-hour volume in August 2026, down from roughly 54% of Solana launchpad share in July 2025. It removed its Buy/Burn and SBR mechanisms in December 2025, which is the clearest signal of how its own team read the trajectory.

Did Vector shut down?

No. Coinbase acquired Vector.fun from Tensor Labs on 21 November 2025 and wound the apps down as part of the integration; the founders and 11 employees joined Coinbase, and the technology went into Coinbase's DEX trading. Lists that count Vector as a failed launchpad have mistaken an acquisition for a collapse.

How can I check whether a launchpad is still alive myself?

Look at 30-day protocol fees rather than the website or the token price, and read the quarterly trend rather than a single day — 24-hour figures lag and misreport. Zero fees over 30 days means no trading on the venue's contracts. A missing protocol page is a stronger signal, though not a proof, since indexing coverage is incomplete.

---

Lock your liquidity before you ask anyone to trust it

The pattern running through every collapse on this page is that off-chain promises evaporated while on-chain commitments did not. Team Finance — built by TrustSwap, which also builds Meme Central — locks LP tokens for a fixed term on Ethereum, Robinhood Chain, Polygon, Base and BNB, and the lock shows as a verified badge on the token's page in the Meme Central feed. It survives the launchpad going dark, which is the point. It does not stop a developer selling their own allocation, and it cannot rescue a token whose venue has stopped producing volume.

---

Nothing here is financial, legal or tax advice. Memecoins are extremely high-risk: most lose most of their value, and the majority of tokens launched never reach a decentralised exchange at all. Never spend money you cannot afford to lose entirely. Meme Central does not recommend any specific token. Data described as Meme Central's own reflects tokens indexed by Meme Central and is not whole-market data.