Changelog

What's new

What we've shipped, newest first. For what's coming next, see the roadmap.

    • API keys can be given an expiry date. Learn more
    • Account-wide limits: a spend limit (“stop everything at $X”) that pauses your rentals when it is reached, and a cap on how many rentals run at once. Learn more
    • The rentals list pages through long histories. Learn more
    • The real Green Compute logo across the dashboard.
    • GPU availability is shown as a state (available, almost fully rented, rented out, on request) instead of a count, with a waitlist for when GPUs are rented out. Learn more
    • Sign in with a wallet (Bittensor or EVM): no email needed. Wallet sign-ins are screened against sanctions lists. Learn more
    • A terminal in the browser for every rental. Learn more
    • Payments sent on the wrong subnet are recorded and refunded by support, instead of being lost.
    • Safer crypto pricing: several price sources, with automatic failover between them.
    • Payment reviews come with a reason, and an email when they are decided.
    • An events feed and signed webhooks: know when a rental is ready, fails or stops. Learn more
    • Per-GPU stats and logs for your rentals.
    • An MCP server, so AI assistants and agents can rent, watch and stop GPUs. Learn more
    • An overview of all your rentals in one place.
    • Alerts when GPUs free up.
    • Run a command on many rentals at once.
    • An email when a running rental stops.
    • Accurate availability, with a clear “no capacity” answer instead of a rental that never starts.
    • A pod is marked ready only once you can SSH in.
    • Paused rentals can resume automatically after a top-up.
    • Low-credit emails and a forecast of how long your balance lasts.
    • The alpha price is shown together with its payment bonus.
    • Spending by day, and monthly statements you can download as CSV.
    • Machine-readable error codes in every API error. Learn more
    • Name your rentals.
    • Fleets: keep a number of rentals running, replaced automatically if one stops.
    • SSH host keys for every rental, so you can check you are connecting to the right machine.
    • Restart policies for your start command.
    • Email notifications.
    • A redesigned dashboard.
    • Private models are billed by the hour while they run. Learn more
    • Terms of Service and Privacy Policy pages. Learn more
    • Deploy progress for models: see each step from download to ready.
    • Your models at their own address (name.inference.green-compute.com). Learn more
    • An agent-ready API: scoped keys with limits, idempotency keys, Retry-After on busy answers, maximum runtimes, and a start command and environment for pods. Learn more
    • Docs for humans and agents, llms.txt and a quickstart. Learn more