Private beta — onboarding in cohorts

The open marketplace for AI compute.

Every result verified. Every token priced.

DistAgent runs AI tasks on a network of independent GPUs — cheaper than hosted APIs, verified before you pay. Have idle GPUs? Plug them in and earn.

Try it now — no signup

See it run on a real GPU

Submit a prompt below. It runs on the live network just like a paid task — you just don't get charged.

What do you want AI to do?

Submit a task and it runs on a real GPU node in our network. You only pay for the compute you use.

Need an idea? Try one of these:

How it works

Three steps from prompt to verified result.

  1. 01

    Submit a task

    Send a prompt or job through the web app or API. No infrastructure to manage.

  2. 02

    Runs on a real GPU node

    Your task is dispatched to a real GPU in the network and executed in seconds.

  3. 03

    Verified result + receipt

    Every result is verified before you pay. You see exactly what each task cost, per token.

Why not just use a big cloud?

Hosted AI APIs are convenient and opaque. DistAgent is built on the opposite bet: open supply, verifiable results, transparent prices.

Trust is shown, not asked for.

Verified compute

Results are checked before you pay. Deterministic tasks can be re-run on an independent node and must agree — a broken or fabricated result never gets billed. Hosted clouds ask you to trust them; we show our work.

Marketplace pricing

Prices come from competing supply — idle GPUs bidding for work — not datacenter margins. Every task comes with a per-token receipt showing the model, the node, the compute time, and the exact cost.

No infrastructure

No instances to rent, no clusters to babysit, no monthly commitment. Submit a task from the web app or API and get a verified result back. Prepaid credits, spend only what you use.

For GPU owners

Earn by sharing your GPU

Have a spare GPU? Download the desktop app and click Start — it sets itself up, no terminal and nothing to configure. Your machine serves tasks from the network while it would otherwise sit idle, and the app shows exactly what each task earned.

DistAgentOnline — serving the network
Models you serve
Llama 3.2 1BServing
Qwen2.5 0.5B InstructServing
Qwen2.5 7B InstructPaused
Tasks served today31
Earned today$0.87
For media & entertainment

Your render farm is idle between shows.

Studios already own exactly the hardware this network runs on. Run DistAgent as a private cluster on your own machines — your content never leaves your network — or earn from capacity you aren't using.

Questions people ask us

Where do my tasks actually run?+

On real GPUs owned by independent providers in the network — from spare workstations to dedicated rigs. Tasks are only dispatched to nodes that advertise they can run them, and every result is verified before you are charged.

Is my data safe?+

Tasks run on independent provider hardware, so during beta we recommend not submitting confidential material. For teams that need hard guarantees, we offer private network deployments — your own hardware, your own network, nothing leaves your control. That is the model we pitch to studios.

What workloads are supported today?+

LLM inference (text generation) on a catalog of open models, honestly labeled. The execution layer is pluggable: image generation and containerized GPU jobs are the next backends on the roadmap.

What does it cost?+

Per token, prepaid. You buy credits, and each task shows its exact cost on a receipt before and after it runs. No subscription, no minimums. Beta participants get starter credits to experiment with.

What does beta access include?+

Full access to the web app and API, starter credits, and the node agent if you want to earn with your own GPUs. We onboard in small cohorts as network capacity grows, so early signups get in first.

Be first in line

We onboard in small cohorts as network capacity grows. Tell us who you are and we'll save your spot.