FAQ: Testnet & Launch Timeline

What exactly is the timeline — when do we spin up?

The project and the testnet were announced on stage at Token2049 Singapore in the first week of October 2026, with the Flop launch event on 6 October; the team was also at Korea Blockchain Week at the end of September.
Devnet comes first and runs in two stages. A private internal devnet is running now to stabilise the network and produce the setup runbook, after which node images are published and the node code is split into its own repository so outsiders can run it. A first batch of institutional miners and validators is being admitted to that private devnet from the week of 13 October — miners and validators first, agents later — with the group widened through the second half of October. Devnet earns nothing: it is a throwaway network for proving setup, with no rewards and no airdrop weight, and the only reason to join is to have machines proven and live on testnet day one.
Testnet: the working target is Tuesday 3 November 2026, running about 90 days. It is a target rather than a fixed date and will be confirmed as devnet onboarding completes. Expect forced software upgrades through the 90 days, each with a short deadline — keeping up with them is part of what the validator ranking measures. If you are timing a GPU lease or a vendor commitment against it, say which date you need to hold and we will confirm against that date rather than leave you to guess. If you want machines on devnet, bring them earlier; otherwise 3 November is the date to plan for. No cap on miners; validators are selected top-1,000 on testnet performance at cutover.
Mainnet, TGE and token listing: Q1 2027, tracking to early February. That is the working date rather than a marketing range, and it comes out of the arithmetic rather than a preference — a testnet starting on 3 November and running ninety days finishes at the start of February, so genesis around early February is the earliest the current schedule supports and anything earlier would mean shortening testnet. TGE happens at the same time as mainnet, not before it.
Campaign and community rewards promised before launch — AMA prizes, competition prizes — are distributed after TGE, against wallet addresses collected in advance. They are not paid during testnet.
Supporting materials: miner and validator onboarding manuals and specifications in the coming weeks; an updated full whitepaper (the earlier draft is superseded — the teaser at flop.finance is the current reference); yellow paper with the full technical spec alongside it.
The earliest date on which you can actually run a node is the private devnet, which is admitting its first external miners and validators now; the first date on which anyone can earn is testnet, targeted for 3 November. There is no earlier start.
You do not need to join testnet on day one, but earlier participation means a larger share of the low-supply months and therefore a larger airdrop.

Is participation open or permissioned? Is there a cap, does the testnet airdrop hold up if compute supply floods in, and does easy rental produce a land rush?

Testnet is open — no whitelist, no restrictions, no cap on miners. As many people who want to mine can mine; there is only so much block reward to go around, and that competition is intended.
There is no application process. A miner and validator form exists and it is optional — it is how we reach you with setup instructions, not a gate, and nobody is approved or rejected on it. That is the same for a single consumer card and for a fund bringing a thousand machines: the rules do not change with the size of the fleet, and there are no side arrangements for anyone.
If you have submitted the form and heard nothing, nothing has gone wrong: it is a contact list, not a queue, and nobody is waiting on an approval. Devnet and setup details go to everyone on it after the 6 October launch.
Validator entry during testnet is equally open — no whitelist, no cap, nothing to buy: the network sponsors the testnet stake. What decides the outcome is a score across uptime, block production, accuracy, latency and geographic spread, measured across the whole three months; no weighting between the criteria is published. Validators selected at cutover are each granted the 1,200,000 $FLOP bond free rather than receiving a foundation delegation; the set is 1,000 at genesis and stays capped at 1,000. After cutover the set is maintained by rotation roughly every 30 days, with about 50 ejected and 50 promoted from a queue of validators in waiting, and committee membership additionally requires an accepted verification duty within the previous 24 hours. The specification also carries a limit of five validator slots per operating entity and a proposed 10% per-provider cap on the set, neither enforced yet; both are being reconciled with the institutional unit guidance.
On how that standard is set and who sets it: the rubric is published in advance and applied without discretion. Nobody is hand-picked and there is no approval step. Infrastructure and geographic diversity, including distinct hosting providers and autonomous systems, form part of the score so that the validator set does not share a single point of failure. The full cutover rubric is still being finalised and is not yet published; it will be published before testnet ends, so anyone spending money to compete can see the standard before committing.
Testnet is deliberately not gated to favour early contributors. Gating would be the obvious way to reward the people who showed up first, and it was considered and rejected: demand is already well beyond what a gate would be protecting, and a permissioned testnet is the wrong signal for a network whose whole claim is permissionless compute.
A large operator is treated like any other miner: its share of the miner airdrop equals its share of verified compute delivered across the testnet. For scale, one Vera Rubin NVL72 rack is about 444 H100-equivalents, which on a testnet of roughly 1,000 units would be close to a third of all compute. There is no per-operator limit on the testnet airdrop in the current design, and no reduced weighting for size; whether a limit is introduced has not yet been decided, and it would be announced before it was. Concentration is handled through verification rather than through a cap — buyers choose their miners, and every job a large operator serves is verified the same way as everyone else’s — and miner concentration does not weaken the network’s security: miners perform work, validators verify and settle it, and the chain’s security rests on the validator set, not on how compute ownership is distributed. Ownership of miners is not tracked.
Expected supply: about 1,000 H100-equivalents by end of testnet, about 1,200 at mainnet start. The miner airdrop pool is fixed, so per-miner airdrop scales with each miner's share of compute delivered — more supply means a smaller share per unit, which is the same dynamic as the block reward. If far more compute arrived — say 10,000 GPUs — each unit’s share of the 1.2 billion pool and of the 96 $FLOP block would fall in proportion; the planning assumption is about 1,000 H100-equivalents, so size your return against your expected share of total network compute rather than against a per-GPU figure. Compute arrives at that scale only when demand, and the market capitalisation that follows it, is there to pull it in. Rewards are earned period by period, so what a fleet earned while supply was thin is not diluted by capacity that arrives later; the advantage is in joining early, not in waiting to see who else comes.

How hard is setup, really? What is the process to get ready before testnet, and what are you offering someone who is not a Web3 person?

One afternoon end to end, and you can rehearse it today without waiting.
1. Rent a GPU from any NeoCloud platform; they hand you root access to a hosted machine. Live capacity, published rental benchmarks and provider-by-GPU pricing are at gpus.flop.finance. 2. Clean install of any major Linux distribution — about 30 minutes. 3. SSH in, install the Flop miner or validator software, plus tuning — one to two hours.
The manual will be a Markdown file written for your own AI to execute rather than a distilled human guide: point Claude or any capable agent at it and have it run the setup, including building a Docker image and running it on a VPS. It is not yet published and no date has been set.
After that everything is automatic — the software loads models per validator assignment, serves inference and accounts rewards.
It does not matter whether you are in Web3. If you can buy or rent GPUs and run the software, you can mine. Apply on the miner form at flop.finance and Flop will reach out with the detailed specifications.
Operational burden is comparable to running a validator node on any other chain. Practical advice: rent one unit now and do a dry run, so internal approval timing does not cost you the early low-supply months.
Setup guides and manuals are being released over the coming weeks.
For an agent rather than a miner, the day-one requirement is smaller still: an identity key (the same DID key used on TechnoCore), the ability to make HTTP requests, and the ability to call the testnet faucet and submit an inference request signed with that key. No wallet software, chain client or SDK beyond that is needed to start; connectors for the common agent frameworks follow.

Is there a devnet before the testnet? Who can join, at what size, and when? Do institutional partners need to fill in the public application form?

Yes. There are two devnets and they are not the same thing. An internal devnet is running now to stabilise the network and produce the setup runbook. External partners are now being admitted to it in small batches, miners and validators first, from the week of 13 October — roughly three weeks before the 3 November testnet target — so partners have machines live and configured on testnet day one. It currently runs about a dozen validators and is being grown toward roughly fifty across regions before testnet.
It is deliberately small: a few dozen institutions. That is why technical support is far more available on it than it will be once testnet opens to everyone.
Devnet earns nothing. No rewards, no airdrop weight, no scoring. Joining is recommended rather than required, and it is not a qualification step for the airdrop — the point is that a fleet is proven before scoring starts, instead of losing the first days of testnet to setup.
On size: one device is enough to validate the setup end to end, and a single pre-access unit from a hardware vendor is a perfectly good way to do physical diligence. Bringing roughly ten GPUs toward the end of devnet is the better test, because that is what proves the configuration holds at scale rather than on one box. There is no advantage to contracting a large fleet early.
Hardware arriving in the second half of October still works. A fleet landing in late October has one to two weeks of devnet before the 3 November testnet target, which is enough to prove setup.
There is no application process and no form to clear, for institutional partners or for anyone else. The public form is optional and exists to deliver setup instructions.
For an operator who wants in early as a validator: there is nothing to apply for and no seats are allocated before testnet. The useful preparation is operational — regions, hardware per node (no GPU; disk for the model set), how many nodes you intend to run (the published spec sets a self-declared limit of five validator slots per operating entity) and prior validator experience — so that machines are proven on the devnet in October and live on testnet day one. The validator set is selected on testnet performance, not on who joined first.
What you get on it: the network to run against ahead of the crowd. Admitted partners receive the join instructions — the miner image, the network access key and, for validators, the node build and chain spec — which go to admitted partners as one package once the internal devnet test completes — none has been issued yet; your machine is then visible on the devnet explorer, so a validator or miner can be running before anyone else. There is no public node repository yet; the public yellow paper on GitHub is the open material. Agents can already reach the devnet endpoints, rate-limited. Stake on devnet, as on testnet, is sponsored by the network.
Public timeline around it: Token2049 Singapore announcement in the first week of October, Flop event 6 October, testnet targeted for 3 November for about ninety days, mainnet and TGE Q1 2027 tracking to early February.

I am not a miner and not a creator — I just invest in crypto. Can I participate in the airdrop, and what exactly do I have to do?

Yes. You do not need hardware and you do not need an audience.
1. Create an agent and get a DID key — the community has built easy ways to do this. 2. When testnet launches, claim testnet $FLOP from the faucet into your wallet (the claim period — hourly or daily — is not yet decided). 3. Instruct your agent to actually spend it on an inference request; the task itself does not matter. 4. Spent testnet $FLOP becomes your claim on mainnet $FLOP. 5. The mainnet allocation is locked, and unlocks as you spend it on mainnet inference at 3:1 — every three $FLOP you spend on inference releases one locked $FLOP. There is no time-based vesting: what you never spend stays locked.
If you don't use it, you lose it. Creating DIDs earns nothing; posting your DID in a TechnoCore room earns nothing.
None of this is live until testnet opens — the working target is 3 November; today you can only do step one.

Is there a Discord or a central place where people can go to learn more about Flop? Where do I get the detailed information?

There is no Discord yet. Official information comes from flop.finance and the @flop_labs X account. The teaser is published at flop.finance.
Applicants to the miner, validator and creator programmes are contacted by email with access to the intro site, which carries more detail on the protocol, and — for creators — the engagement homework.
The published teaser (flop.finance/whitepaper) carries the current airdrop, testnet and participation rules and the recommended hardware section, and the public yellow paper (github.com/flop-labs/yellowpaper) is the normative protocol specification. A fuller information pack for prospective miners and validators, with an expanded airdrop and testnet section, is being published ahead of testnet; the full whitepaper is being revised and is not a current source.
If you see a Flop token trading anywhere, or a site offering a mint or a claim, it is a scam — the token does not exist yet.

Are US persons — or participants from any particular country — eligible to mine, validate, receive the genesis airdrop and earn testnet rewards? Can you confirm there are no jurisdictional restrictions before we commit hardware?

Flop Labs develops open-source software. The protocol is code that is publicly deployed; there is no operator, no admissions step and nobody in a position to admit or exclude a participant.
For that reason we do not issue eligibility statements for any country or category of person, and we do not confirm that any particular participant is permitted to take part. There is no entity that could grant that permission, and we will not give the impression that one exists.
Each participant has to determine for themselves, with their own advisers, whether they can supply compute to, validate on, or receive tokens from an open-source network under the rules that apply to them.
What we can offer is the code itself and the published specification, which describe exactly how mining, validation and the genesis allocation work. If that is not enough for a participant to make the decision, the right course is not to join the network.
On node location specifically: the protocol has no restriction on where a miner or validator runs and no IP-based gating. Demand comes from the US, Europe, the Middle East and Asia, so a geographic spread of nodes — roughly even across Europe, North America and Asia — is what the network wants, and the market, not Flop, sets the split.
The same applies to mainland China: the protocol does not block any location, and whether a participant there can run a node is a question for that participant under its own law, exactly as it is for bitcoin mining. Flop will not interfere either way and will not advise on it.
The same applies to a business integrating Flop into its own product: Flop is an open, neutral network, the risk assessment depends on where the integrator is based and what it does, and it is a decision for the integrator and its own lawyers. Flop does not assess it for them and does not restrict participation by region — the network does not care where a miner, validator or agent is from.

Is there a devnet token? Are testnet $FLOP balances transferable — could a miner accumulate testnet tokens and pass them to agents so they use more inference before mainnet? Can our agent route inference to Flop during devnet or testnet?

There is no devnet token. Devnet is a throwaway network for proving setup: no rewards, no airdrop weight and nothing that carries forward.
Testnet runs its own $FLOP-equivalent play token so that paying for inference, earning and staking behave as they will on mainnet. It has no value and does not convert into mainnet $FLOP: what is owed at genesis is computed from vouchers that record verified work up to a published snapshot block height, not from testnet balances, so moving testnet tokens between accounts does not change anyone's airdrop.
What can be transferred before mainnet is the voucher that records verified work — a token on the Flop chain that can be held and moved like any other balance and sold peer-to-peer through the network's HTLC atomic swap, with the unlock obligation travelling to the buyer, but that cannot be spent on inference — not the testnet $FLOP balance itself. Whether testnet balances can be moved between accounts is an implementation detail that has not been finalised, and it does not matter for the airdrop: moving testnet tokens changes no one's allocation. Agents draw testnet $FLOP from a faucet with per-identity limits, so an agent's inference volume on testnet is bounded by the faucet rather than by how much it can acquire.
Routing an agent's inference to Flop is exactly what testnet is for and is welcome. It is not possible on devnet, which is closed to a small number of institutional partners and exists to prove node setup rather than to serve demand. On mainnet agents buy compute with real $FLOP and there is no limit on how much any agent can use.
Miners earn real $FLOP on mainnet and nothing is held back; the arrangement above applies only to the ninety-day testnet.