voxlm

Coming soon · Peer-to-peer AI

AI that belongs to the people who run it.

VoxLM is a peer-to-peer network where anyone can contribute compute, data and feedback to build and run language models together. Like BitTorrent, but for intelligence.

No spam. One email when the first node ships.

You're in. Watch your inbox.

That didn't go through. Check the address and try again.

Why

Today's AI runs in a handful of data centers owned by a handful of companies.

  1. 01

    Centralized.

    Whoever owns the compute writes the rules.

  2. 02

    Expensive.

    You rent intelligence by the token, forever.

  3. 03

    Fragile.

    One outage, one policy change, one acquisition, and it is gone.

How it works

A network you can join from your own machine.

  1. 01

    Install a node.

    A small app for macOS, Windows and Linux. Runs quietly in the background.

  2. 02

    Share what you choose.

    Idle GPU or CPU time, disk space, curated data or feedback. You set the limits.

  3. 03

    The swarm does the work.

    Models and workloads are designed to be split across many peers, so no single machine, and no single company, holds the whole thing.

  4. 04

    Contributors come first.

    Share more, get priority access and a say in what gets built next.

All of this is in development. Details will change.

Principles

Three things we will not compromise on.

  • Open by default.

    Open node software, open process, open discussion.

  • Resilient by design.

    No single point of failure. If one peer leaves, the network keeps going.

  • Owned by contributors.

    The people who provide the compute and the data decide the direction.

The analogy

BitTorrent proved that millions of strangers can share files without a central server.

VoxLM applies the same idea to intelligence: sharing compute, models and knowledge across a swarm.

clientclientclientclientclientclientclientclientserver

One server, many clients

Everything flows through one owner. Remove it and nothing works.

peerpeerpeerpeerpeerpeerpeerpeerpeerpeer

Many peers, many peers

Every peer gives and takes. Remove one and the rest carry on.

Roadmap

Where this is going. No dates, on purpose.

  1. Now

    Waitlist and community

    In progress

  2. Next

    Node alpha, share compute

    Planned

  3. Then

    Distributed inference across the swarm

    Planned

  4. Later

    Collaborative fine-tuning and training

    Planned

  5. Later

    Contributor governance

    Planned

FAQ

Questions people ask first.

Is VoxLM a cryptocurrency?

No. There is no token. Contribution is planned to earn priority access, and the long-term model is still being designed.

What can I contribute?

Idle compute, storage, data you have the right to share, and feedback on model outputs. You choose, and you can stop at any time.

Is my data safe?

You decide what to share. The network is designed so that no single peer holds a whole model or dataset. Full details will be published with the node software.

Is it open source?

The node software is planned to be open source.

When can I try it?

We are building in the open. Join the waitlist and you will get one email when the first node is ready.

Who is behind this?

An independent team. More soon.

Be in the first swarm.

Early contributors shape what gets built. Leave your email and we will tell you when the first node is ready.

No spam. One email when the first node ships.

You're in. Watch your inbox.

That didn't go through. Check the address and try again.