Introducing Halo: intelligence that belongs to everyone

We taught machines to reason, write, plan, and build alongside us — and then, almost in the same breath, we started walling that ability off. The most capable models now sit inside a few companies. Access is metered and rate-limited, locked to some regions and denied to others, and it can be revoked without warning. Entire countries are shut out. Entire use cases are quietly ruled out. The terms shift overnight, and you usually learn about it the moment the thing you relied on goes quiet.

That’s a strange fate for the most consequential technology of our era: gatekept and permissioned before it ever reaches most of the people it could help.

We think there’s a better answer. Intelligence shouldn’t be gatekept, controlled, or permissioned by anyone — it belongs to everyone, or it belongs to no one. We chose everyone. Today we’re releasing Halo, an open marketplace for intelligence that no single company owns.

What Halo is

Halo is a peer-to-peer network for AI inference. If you need a model, you can reach one. If you have access to a model, you can serve it. Money moves directly between the two sides in stablecoins — no platform sitting in the middle deciding who’s allowed to participate.

There are three ways to take part:

• As a consumer — a person who just wants to prompt a model. Open the Halo app, add funds, and go.
• As an agent — autonomous software that needs inference to do its job. Agents are first-class here: give one a wallet, a budget, and a local OpenAI-compatible, censorship-resistant endpoint, and it can pay for its own thinking with no human in the loop.
• As an operator — anyone who serves inference and earns for it.

This is a public alpha. It works end to end today, on Base mainnet, with real USDC. It’s also genuinely early — there are rough edges, fixes in flight, and features still landing. We’re building it in the open on purpose: a network for intelligence should be built in public, with the people who are going to use it.

Onboarding: deposit, sign once, prompt

The whole system rests on two principles — you pay for what you use, and you never ask permission.

To begin as a consumer or an agent, you deposit USDC on Base into your Halo balance, along with a small amount of ETH on Base. Nearly everything after that is gas-sponsored, so you don’t pay network fees to run inference. The single transaction you sign yourself is that first deposit — which is the only reason you need a bit of ETH to start. From then on, prompts draw down from your balance automatically, and you can top up or withdraw whenever you like. Your wallet is your identity; your balance is your limit. Nothing else stands in the way.

That’s the entire flow: deposit, sign once, prompt.

For operators: turn model access into income

If you can reach a model, you can serve it on Halo. Operators run a single command and connect to the network. Serve anything you have access to — a frontier model behind an API key, an open model you host yourself, or a local model on hardware you already own. The network routes paying requests to you, and you earn USDC for each one. No staking, no token to buy, no gatekeeper judging whether you qualify.

This is what lets Halo route around control. When access to intelligence is peer-to-peer, a block in one place doesn’t cut everyone off. If a model isn’t available where you are, someone else on the network can still serve it to you. There’s no single front door for anyone to close.

What Bitcoin did for money, Halo does for intelligence. Bitcoin made value something you can hold and send without anyone’s blessing. Halo makes intelligence something you can reach and provide the same way — openly, directly, and beyond the reach of any one institution.

Why we center operators, not idle laptops

Plenty of mesh projects have tried to crowdsource AI from spare cycles on random consumer devices. The intent is good; the outcome is usually unusable — slow, unreliable, and limited to tiny models nobody really wants. Pooling leftover compute doesn’t get you the models people actually need.

Halo takes a different stance. We put the operator at the center: someone who deliberately provides real, capable models and runs them well. Idle compute is welcome, but the network is designed so the best models can be served reliably without a central host — not so you settle for whatever happens to be lying around. The goal isn’t “some AI from everywhere.” It’s the models you’d actually choose, available without a gatekeeper.

Privacy you don’t have to trust us for

For sensitive work, Halo supports fully confidential compute through NEAR TEEs. Your prompt is handled inside a trusted execution environment, encrypted even while it’s being processed, where it can’t be read by the operator, by us, or by anyone else along the way. You don’t have to trust the network with your data — the hardware enforces it, and your browser can verify the proof. Private intelligence shouldn’t be a premium tier reserved for enterprises. On Halo, it’s part of the network.

Founding Inference Contributors

A marketplace for intelligence needs intelligence on day one. We’re grateful to our Founding Inference Contributors — Venice.AI, NEAR, and 0G — who are providing capacity to bootstrap Halo so there’s real model availability the moment it opens. Their support means the network launches with capacity, not just the promise of it. Thank you for backing an open foundation for AI.

What’s next

This is the starting line, not the finish. In the coming days we’ll publish the full Halo roadmap, including SPEX (verifiable inference) — statistical proof that an operator actually ran the model you asked for — along with federated relayers, image generation, and more. We’ll keep shipping in the open.

Try it

The alpha is live. Add funds and prompt something. Bring a model and become an operator. Point an agent at it and let it pay its own way. And tell us what breaks.

We built Halo because who controls intelligence is one of the defining questions of this century — and the answer can’t be “a few companies.” It has to be everyone, or it’s no one.

We chose everyone. Come build it with us.