FLock.io

About FLock.io (FLOCK)

What Is FLock.io (FLOCK)?

FLock.io is a decentralized network for training, fine-tuning, and monetizing AI models. Rather than one company owning the data, the compute, and the finished model weights, FLock splits those jobs across independent participants who stake the FLOCK token, do measurable work, and are paid or penalized based on how that work scores.

In practice the network is a competitive training platform, a federated learning layer that lets a model improve on private data without that data being pooled, and a hosting layer where finished models are served through an API and can be turned into on-chain assets. FLOCK is the network's token: it is the stake required to participate, the unit rewards are paid in, and the governance token of the project's DAO.

Official sources:

• FLock.io
• FLock documentation
• FLock whitepaper
• FLock.io on GitHub
• FLOCK token on BaseScan

Who Founded FLock.io? (History)

FLock.io was founded by Jiahao Sun and is based in London, with a research-heavy team that includes computer science academics from the University of Oxford. The incentive design is not purely a product idea: the mechanism used in the federated learning layer follows the team's peer-reviewed work published in IEEE Transactions on Artificial Intelligence, which describes combining a federated learning system with a blockchain to defend against poisoning attacks without relying on heavy cryptographic protocols.

The project raised a 6 million dollar seed round announced in March 2024, co-led by Lightspeed Faction and Tagus Capital with participation from Digital Currency Group, OKX Ventures, and Volt Capital, followed by a 3 million dollar strategic round in December 2024 led by Digital Currency Group with Animoca Brands, Fenbushi Capital, and others.

FLock's mainnet launch and token generation event took place on Base on January 3, 2025, bringing AI Arena, FL Alliance, and the AI Marketplace live together with the FLOCK token.

How FLock.io Works

FLock describes its system as a pipeline a model passes through, with a different set of incentives at each stage.

1. Competitive training in AI Arena A task defines the model that is wanted and the validation data it will be judged on. Training nodes fine-tune or build candidate models and submit them. Validators score the submissions and reach consensus on those scores, which produces a ranking and, from it, a consensus model. Training nodes and validators must stake FLOCK to take part, and both can be delegated to by holders who do not want to run the work themselves.

2. Federated refinement in FL Alliance The consensus model can then be handed to federated learning clients who improve it using their own local data. Roles are assigned using random functions, participants stake to commit, and a slash-and-reward mechanism penalizes clients whose updates look malicious while paying the honest majority. The raw private data never leaves the participant.

3. Deployment and monetization The resulting model is hosted and served, historically through the AI Marketplace and now through the FLock API Platform, where real inference calls generate revenue. FOMO, the FLock Open Model Offering, is the layer that turns a specific model deployment into an economic asset funded in FLOCK and backed by that inference usage.

Key Ecosystem Components

The parts of the network a reader is most likely to encounter:

• AI Arena, the competitive training platform reached through train.flock.io, where task creators, training nodes, validators, and delegators interact
• FL Alliance, the federated learning layer for improving models on proprietary data held by different parties
• AI Marketplace, the original hosting and deployment surface for trained models
• FLock API Platform, the inference layer where models are served in production and calls are paid for
• FOMO, a launch and revenue-routing protocol in which model deployments are funded through Real Model Asset offerings and represented by fixed-supply Model Tokens alongside FLOCK
• The FLock DAO, through which token holders propose and vote on protocol and treasury decisions

FLOCK Token Utility

FLOCK is used for several distinct purposes across the network:

• Staking, which is the precondition for acting as a training node, validator, or federated learning client, and which is what gets slashed for dishonest behavior
• Delegation, where holders back a specific training node or validator and share in that participant's rewards according to a ratio the participant sets, subject to local availability restrictions
• Rewards, since contributions across AI Arena and FL Alliance are paid out in FLOCK
• Payment and access, including fees to join the system, which also raise the cost of denial-of-service style spam, and spending to reach hosted models
• Governance, where submitting a proposal requires the proposer to stake tokens first as a filter against frivolous or malicious proposals
• Funding model launches in FOMO, where issuers raise FLOCK to fund a concrete deployment and tokens are minted only if the launch clears predefined economic thresholds

Total supply of FLOCK is capped at 1 billion tokens. Base is the token's home network and the default network for FOMO, and FLOCK is also deployed on BNB Chain, Ethereum, HyperEVM, and Solana.

What Makes FLock.io Unique?

The distinguishing choice is that FLock treats model quality as something to be measured and scored rather than asserted. A training node does not get paid for submitting a model; it gets paid if validators, who are kept unaware of which submission came from whom, score that model well against held-out data. That single decision drives most of the rest of the system, including the staking requirement, the slashing rules, and the delegation market that forms around consistent performers.

The second distinguishing choice is that private data is treated as a first-class input. FL Alliance exists so that a hospital, a bank, or an enterprise can improve a model without handing over its dataset, which is a materially different proposition from platforms that only coordinate compute.

How FLock.io Differs from Similar Networks

Compared with other decentralized AI networks, FLock is positioned less as a compute marketplace and more as a training and evaluation network:

• The unit of work is a trained model judged against a validation set, not raw GPU hours rented out
• Rewards concentrate on top performers by design, rather than being spread evenly across all participants
• Federated learning is built in as a distinct stage, which is uncommon among crypto AI projects
• Monetization is tied to real inference consumption through the API Platform and FOMO rather than to emissions alone

Decentralized compute marketplaces and inference networks solve an adjacent problem, and in practice can be complementary to what FLock does rather than direct substitutes.

Use Cases of FLock.io

Typical uses of the network include:

• Fine-tuning open models for a specific domain through competitive tasks
• Improving a model across several data holders who cannot legally or commercially share their data
• Hosting a finished model and serving production inference through an API
• Funding a model deployment through a FOMO offering and sharing in the revenue it generates
• Staking or delegating FLOCK to earn a share of training and validation rewards
• Governing protocol parameters and treasury decisions through the DAO

How FLock.io Is Secured

Security in FLock is primarily economic rather than purely cryptographic. The documented defenses include:

• A minimum stake for participation, which makes Sybil attacks expensive because every identity has to be funded
• Blinding validators to the origin of submissions, which reduces the scope for collusion between validators and training nodes
• Paying only top-performing training nodes and validators, which removes the payoff from free-riding and low-effort submissions
• Using two distinct evaluation datasets, so optimizing against one does not guarantee a good score and lookup-style gaming is frustrated
• Majority voting plus slashing in the federated learning stage to defend against model poisoning, with part of the slashed stake going to whoever calls the slashing function
• Rate limiting, so a single participant cannot monopolize platform resources

The blockchain layer holds the stakes and automates reward distribution, so payouts and penalties follow the recorded results rather than an operator's discretion.

Risks and Considerations

Points worth weighing before treating FLOCK as a long-term holding:

• Task creation is still centralized. FLock's documentation states that tasks are created by the FLock.io team, with community task creation planned rather than shipped, so the network is not yet permissionless end to end.
• Reward concentration cuts both ways. A design that pays only top performers can leave smaller participants with running costs and no return.
• The economics depend on real inference demand. FOMO explicitly routes value from paid API usage, so if usage stays thin, the model-token layer has little to distribute.
• Delegation is not available everywhere, and the project notes that local law may restrict it.
• Multichain deployment means users must confirm which network a given FLOCK balance sits on before moving it.
• Decentralized AI is a crowded and fast-moving sector, and roadmap items can slip or change shape.

Why FLock.io Matters

FLock is one of the more concrete attempts to answer a real question: if AI model training is valuable, who captures that value, and how do you pay strangers fairly for contributions you cannot fully verify by inspection? Its answer, a staked and scored training market plus a federated layer for data that cannot be shared, is specific enough to be tested rather than merely described, and it is being pushed toward revenue from actual model usage instead of emissions alone.

Where Can You Buy FLock.io (FLOCK)?

FLOCK trades on both centralized exchanges and decentralized venues. On-chain liquidity is concentrated on Base through Aerodrome and Uniswap, with further pools on BNB Chain through PancakeSwap and on Solana through Raydium. Availability, trading pairs, and supported jurisdictions vary by venue.

Major Centralized Exchanges (CEX)

FLOCK has been listed on venues including:

• Kraken
• Coinbase
• Bybit
• Bitget
• Gate
• KuCoin

Listings change over time, so check the exchange directly before assuming a pair is available in your region.

Storage Options After Purchase

After buying, FLOCK can be held:

• On an exchange, where the exchange controls the keys
• In a self-custody EVM wallet for the Base, BNB Chain, Ethereum, or HyperEVM deployments
• In a Solana-compatible wallet for the Solana deployment
• Behind a hardware wallet used with compatible wallet software

Because the same ticker exists on several networks, always confirm the network and the contract address before depositing or withdrawing. Staking and delegation are performed through FLock's own interfaces rather than in a generic wallet.

Related Coins

You can compare FLOCK with related networks on Coinrithm:

• Bittensor (TAO)
• Artificial Superintelligence Alliance (FET)
• Render (RENDER)
• Akash Network (AKT)
• io.net (IO)