Open betalegit2compute · the community AI network
Lend your PC's idle hours to a community network that runs open AI models and media jobs. It steps aside the moment you're back, and every job it finishes is checked and credited to you.
Free to join · sign up with Discord · Windows and Linux · pause any time
Away mode
Away mode is one switch in the legit2node panel, Only run when idle. With it on, your PC takes work only when nobody is using it.
Move the mouse, start typing, launch a game or a full-screen video, turn on Do Not Disturb, or unplug your laptop, and new work stops.
Work resumes once you've been away for the time you choose: ten minutes unless you change it. A locked screen counts as away.
A job that is already running gets a short grace period to finish, so the work your PC started still counts.
Pause for an hour from the legit2node panel whenever you want every bit of your PC to yourself.
Giving back
The network exists because people lend their machines. Here is what that gets you.
No bidding, and no renting your machine out to strangers. The network sends the work, and every job your GPU finishes is metered and credited to you.
Earnings are held until the job's output passes validation, then clear after 14 days. After that they are yours to spend or cash out.
Turn earnings into platform credit and get +25% on top, for clips, voice, music, images and models across legit2. Or cash out by PayPal, bank transfer or crypto (10% fee).
Your tier sets your share of the model serving your GPU runs, and your rate for media jobs. Keep your node reliable to move up; founder rates are hand-picked.
Choose which kinds of jobs to run and how much disk they may use. Nothing downloads until you switch it on, sizes are shown first, and jobs run inside Docker containers the agent manages.
Signing up with Discord puts you in the server where providers ask questions and the team answers. The leaderboard shows who is carrying the network, under a pseudonym.
Get started
One click, and you're in the community server too.
One download for Windows or Linux. It detects your hardware and shows what each kind of job needs before it downloads anything.
Then get on with your day. Your dashboard shows each job and every credit as it lands.
The network today
| # | Provider | Share of work | Jobs | Model requests |
|---|---|---|---|---|
| 1 | contributor-268bcc | 95.7% | 262 | 2,328 |
| 2 | contributor-7c8a2f | 4.3% | 327 | 0 |
Providers appear under a stable pseudonym, never a name or a machine.
Where we are
A year of building, the last few months with a small group of providers and testers. Now it's open to everyone, and still a beta: expect rough edges, and tell us about them in Discord.
One tool with one job: turn long videos into short, captioned clips.
The clipper became a platform that spreads work across many GPUs: clips, voice, music, images, video and open models.
Community providers lent their GPUs and the network ran more than 6,000 jobs, while we learned what it takes to make lending a PC worth it.
Anyone can join with Discord, and one account works across legit2clip, legit2cloud and legit2compute.
For builders
One API key for open chat and vision models, clips, voice, music, images and video. Every job you run is metered to the provider whose GPU ran it.
An OpenAI-compatible endpoint, plus l2c/auto to pick the right model per request. Where the same model is sold elsewhere, ours costs less than the cheapest provider listed on OpenRouter.
Finished jobs so far, by kind.
Want the finished product rather than the API? Make clips on legit2clip, or music, voice and sites on legit2cloud. Same account, same credit.
Run Hermes Agent on our models and talk to it from your server: it answers, uses tools, remembers, and with our skill it makes images, music, voice and video right in the chat. We run our own this way, at about a quarter of a cent per reply.
Questions
Not with away mode on. New work stops the moment you touch the mouse or keyboard, start a game or full-screen video, or unplug your laptop. A job that was already running gets a short grace period to finish.
It depends on how much work the network has and what your machine can run, so we don't promise an hourly rate. Your dashboard shows every job and every credit as it lands.
AI work for the network's customers: transcription, clip rendering, voice, music and image generation, and hosting open models. It runs inside Docker containers the agent manages, and you choose which kinds of work to accept.
You set a budget. Each kind of job shows its download size before anything is pulled, and nothing downloads until you switch it on.
Earnings clear 14 days after the work is validated. Spend them as platform credit with a 25% bonus, or request a payout by PayPal, bank transfer or crypto (10% fee).
No. The agent reports what your hardware can do and the network only sends work it can run. More GPU memory opens up more kinds of work.
Yes. Linux nodes run headless and take work around the clock, without away mode.
Free to join, paused whenever you're using it, and credited for every job it finishes.