Loading
Citrate serves institutions that need compute they can prove, organizations that want to run and build AI on hardware they already own, and the developers building on a network they can verify line by line.
Run inference and training inside your perimeter. Hold your own keys. Train across your sites with federated learning, without exposing raw data. Every result leaves a receipt your auditors can check.
See the institutional posture →Join the network to run your own node, build and deploy AI applications, and use the full toolbox. Your membership stakes a SALT grant that makes you a participant in securing the network. Your data stays on your floor.
Become a member →Drop-in compatible with the OpenAI and Anthropic APIs, so your tools work unchanged. Open specifications, memory-safe Rust, 121+ formal proofs, TypeScript and Python SDKs. Read the architecture, run a node, build on top.
Read the technology →Citrate is an open market for AI work. Independent providers run your inference and training jobs. Every result leaves a receipt you can audit. Payment settles automatically. There are no hourly servers to rent and no lock-in. Anyone can join, every result is proven, and your data stays where you put it.
Every settled job leaves an exportable receipt. Pick the proof you need: a committed fingerprint, a cryptographic proof anyone can check, or a sealed hardware enclave. Dispute a bad result and the protocol slashes the party at fault.
See how verification works →Sealed at rest with AES-256-GCM under a quantum-safe key exchange. Run the entire network on-premises or fully air-gapped. Federated learning lets the lessons travel while the raw data stays in the building.
See how it deploys →On-prem is available now. CMMC and SOC 2 are in remediation. FedRAMP is on a published roadmap. 121+ formal specifications back the substrate underneath.
Read our compliance posture →Business logic, logistics, a shared language for what things are, and proof after the fact of what happened. Citrate is the common ground for all four, across nineteen industries.
Procurement, provenance, and audit substrate for primes and tier-N suppliers.
Federated learning for predictive maintenance and supply-chain audit.
Cross-institution learning without exposing patient data.
Federated fraud models and SR 26-2 aligned model risk audit.
Give students AI literacy on models you own and run, on your own hardware. FERPA-grade by design.
Offer sovereign compute and AI tools to your community, on hardware you already maintain.
Run AI on your own data and build on the network, without shipping anything to a cloud or standing up a platform team.
Today it does. The compute that trains the models everyone uses sits in a handful of data centers, owned by a handful of companies, paid for with data the rest of us hand over in exchange for chat.
Citrate distributes that compute across the machines that already exist, in schools, offices, homes, and hospitals. The people who own those machines get paid. Their data stays on their floors. The public ledger is open enough that a regulator can check it and find nothing hidden.
That is the network we are building: the work, and the reward for doing it, distributed across the institutions and people who do it.
These numbers are checked into source and reproducible. The verification packet lists how to confirm each one.
Look around free. When you're ready, membership is $48 a year, and the first conversation costs nothing.