Sign in to access your compute dashboard, or create an account.
We’ve sent a confirmation email to . Click the link in that email to activate your account.
The email may take a minute or two to arrive. Check your spam folder if you don’t see it — the sender is hello@opentoken.global.
Choose a password to finish setting up your account.
Tell us a bit about yourself so we can show you the right dashboard.
Complete your profile to help us understand your compute needs and context. This is required before you can submit a project.
Resources you submit appear on the public Impact page, alongside your project, once reviewed by the OpenToken team. Each output you contribute compounds the value of the Collective for everyone working under similar constraints.
Tell the story of what your project has achieved — a short summary, headline numbers, and an optional report document (PDF). You can add or update this at any time, while the project is running or after it’s finished. The OpenToken team reviews it before it appears publicly on the Impact page; once live, your updates show straight away.
Track your contributions and see the impact of your support.
These projects have been vetted and triaged by OpenToken. Select a project to allocate compute.
Register your GPU infrastructure to make it available for allocation.
The more you fill in here, the faster we can match this capacity to a project. Required fields are marked with *.
Light-touch settings for your provider account. The allocation preference here applies to every compute source you register — you don’t need to set it again per source.
How would you like compute you contribute to be allocated? You can change this later.
Manage projects, compute sources, usage, and donations.
For when project info has come in by email and you want to put it into the system so the innovator doesn’t have to start from scratch. Creates an auth user, a profile, and (if you fill in the project section) a pre-populated project at the submitted stage.
Each researcher’s compute instance runs a tiny script that POSTs nvidia-smi
readings every 5 minutes so we can see actual GPU utilisation on the pipeline.
Issue a project a token here, then SSH into the box and paste the install command.
Tokens are shown once — copy before closing the panel.
Each row is a project that has a box-agent token issued. Last seen is the most recent telemetry POST. If a box that should be reporting hasn’t checked in for over an hour, something’s wrong (cron stopped, token rotated without re-running the installer, network egress blocked).
The workflow: an innovator submits an output from their My Outputs panel → it lands in Pending review below, attributed to whoever submitted it → you Publish it (live on the project’s entry on the public Impact page), Request changes (the submitter sees your note), or Reject (remove it). You can also add an output yourself with Add an output — useful when work is already public (e.g. a Hugging Face collection) and you want it surfaced without waiting for a submission.
Add an output and attribute it to a project — useful when an innovator’s work is already published (e.g. a Hugging Face collection) and you want it on the Impact / Capabilities surfaces without waiting for them to submit it.
What each completed project promised to deliver, and what’s arrived. Every completed project owes an impact report plus the outputs it committed to at submission. Tick items off as they land — or use Add an output above to publish them.
The stewardship agent reviews telemetry daily and queues anything that needs attention. Researcher-facing drafts are never sent automatically — review, edit if needed, and send from here.
scripts/box-agent/audit.sh on the box
to confirm what work has actually been done.