Skip to main content
Darkbloom runs AI inference, GitHub Actions jobs, and interactive Linux and macOS sandboxes on Apple silicon Macs owned by people in its network. You use one Darkbloom account and one prepaid balance for all of them, and every Mac that does the work proves with hardware-backed attestation which Darkbloom build and settings it is running.

Inference

For developers who want to call a model from their own code through an OpenAI-compatible API and pay per token.

CI

For teams whose GitHub Actions jobs are slow, or that need macOS runners, and want each job to bill the same account.

Sandboxes

For you or your coding agent to SSH into a Linux or macOS VM that keeps its workspace between boots.

Earn with your Mac

For people with an Apple silicon Mac who want to serve inference, host CI jobs and sandboxes, or both.
Prefer your own hardware? Run models on your own Mac with a local OpenAI-compatible API, and send only the overflow to the network.

How the pieces fit

  • One account. Sign in once with your Darkbloom account, and use the same account on every console.
  • One balance. Add funds at console.darkbloom.dev, and inference, CI and sandboxes all draw from that balance.
  • Attested Macs. Each Mac proves what software it runs with hardware-backed attestation, and Darkbloom checks that proof before it sends the Mac a request, a job or a sandbox. See security and trust for what each product does and does not claim.
New here? Choose a product in a minute, or read accounts and consoles to see where keys and funds live. Pricing covers how each product is billed.

Available models

Inference serves the models that are online now, and the catalog changes as Macs join and leave. GET /v1/models lists the models you can call, and their prices are public at /v1/pricing. See models for how to choose one.
Inference and agent sandboxes are experimental and under active development. Expect rough edges and breaking changes.