What OpenGradient built
OpenGradient runs a decentralized network for full-stack verifiable AI: on-chain model hosting, verifiable inference and confidential computing. Developers can choose hardware-protected TEE enclaves or privacy-preserving zkML proofs depending on what their application needs, and build agents and apps through OpenGradient’s Python SDK and CLI.
So far, its Model Hub hosts more than 4,400 models, and the network has processed 33.8B tokens across more than 2 million private inferences, over 1 million of them in the last 30 days. Products built on it include BitQuant, a personal AI quantitative analyst, and DreamFort, a private AI suite for chat, image and video that has served over 1.7 million inferences. OpenGradient is backed by investors including a16z and Coinbase Ventures, and is part of the NVIDIA Inception Program.
Where Lumera comes in
Lumera is building the trust layer for AI agents: what an agent needs to prove who it is, what it was allowed to do, and that its work was actually done. It starts with Lumera ID, live today, which gives people and agents a verifiable on-chain identity. When an agent acts, Lumera ID answers the questions verifiable compute leaves open: which agent acted, and who stands behind it.
Next on Lumera’s roadmap are verifiable execution, so a job can be confirmed as done before anyone pays for it, and settlement that releases payment only on confirmed completion. Together, those are designed to give each agent a track record it can carry from one counterparty to the next.
Why it matters for the ecosystem
AI agents are starting to trade, pay and make decisions on people’s behalf. For that to be safe, three things need to be checkable: that the model ran as claimed, that the agent is who it says it is, and that it was allowed to act. OpenGradient makes the first one provable at scale. Lumera ID makes the second one provable, and Lumera’s roadmap addresses the third. Together, those checks give builders and enterprises a clearer path to agents they can audit and rely on.
What’s next
This is the start of a relationship between two teams working on trustworthy AI, with verifiable execution and settlement next on Lumera’s roadmap.
Verifiable AI is only as trustworthy as the agents acting on it.
About Lumera Protocol
Lumera Protocol is the Intelligence Layer for autonomous systems, providing the infrastructure AI agents and applications need to operate, transact, and build trust independently.
Lumera brings together five core capabilities in one open protocol: durable memory, digital identity, verifiable execution, economic coordination, and trust. Together, they allow autonomous systems to remember across sessions, prove who they are and what they did, pay and get paid, and build a verifiable track record over time — without depending on any single platform or provider.
Built as a decentralized network, Lumera creates a persistent, verifiable foundation that can follow AI agents and applications across models, platforms, and services. Durable memory is live on mainnet today through Cascade, with identity, execution, coordination, and trust capabilities building on the same infrastructure.
For more information, visit lumera.io and follow Lumera on X, Telegram, and Discord.
Website | Twitter | Discord | Telegram | Github | Reddit | Instagram