Photo: Tima Miroshnichenko / Pexels
The AI inference network's disclosure under Blockworks' Token Transparency Framework scored a perfect completion, a first sign that crypto projects are starting to take standardized reporting seriously.
OpenGradient just passed what amounts to a financial physical for token projects, and the results came back clean. The decentralized AI compute network’s B-1 Token Transparency Filing under Blockworks’ Token Transparency Framework achieved “Complete” status with zero gaps, meaning every required disclosure field was filled and verified.
The filing, published on August 18, 2026 and available on the Bloomberg Terminal, covers everything from team backgrounds and token structure to allocation breakdowns and vesting schedules for the OPG token.
What the filing actually reveals #
OPG has a fixed total supply of 1 billion tokens. The allocation splits paint a picture of a project that’s heavily weighted toward ecosystem growth over insider enrichment, though the math is worth examining closely.
The largest chunk, 40%, goes to the ecosystem. The Foundation and Core Contributors each receive 15%. Investors and advisors collectively hold 10%, while staking rewards account for another 10%. Liquidity and launch provisions take 6%, and an airdrop allocation rounds things out at 4%.
The token’s Initial Token Generation Event took place on April 21, 2026, roughly four months before the filing went public.
The tech underneath: verifiable AI on a blockchain #
OpenGradient isn’t just another token with a whitepaper and a dream. The network runs what it calls a Hybrid AI Compute Architecture, or HACA. In plain terms, it splits the work of running AI models from the work of proving those models actually ran correctly.
Specialized GPU nodes paired with Trusted Execution Environments handle the actual computation. Separate full nodes then verify those results using one of three methods: TEE attestations, zero-knowledge machine learning proofs, or cryptographic signatures. All of this gets settled onchain, creating an auditable trail of every AI inference job.
OpenGradient’s approach lets anyone verify that an AI workload ran as claimed without needing to re-run the entire computation.
The network runs on CometBFT consensus and is EVM-compatible, which means existing Ethereum tooling and smart contracts can interact with it without a complete rewrite. It also supports the x402 protocol for payment-gated LLM inference, essentially allowing developers to charge per-query for large language model access directly through the protocol.
The operational numbers back up the architecture’s viability. OpenGradient has processed over 2 million verifiable inferences and more than 500,000 verified cryptographic proofs as of the filing period.
Backing and competitive positioning #
OpenGradient raised $9.5 million from investors including a16z crypto and Coinbase Ventures.
The transparency filing itself matters beyond OpenGradient’s individual story. Blockworks’ Token Transparency Framework is still relatively young, and each high-profile completion builds the case that standardized disclosures should be table stakes for any serious token project.
For investors evaluating the OPG token, the filing removes one layer of uncertainty. It doesn’t guarantee returns or eliminate smart contract risk, but it does mean that basic questions about supply, allocation, and team structure have documented answers sitting on a Bloomberg Terminal rather than buried in a Discord FAQ. Disclosure: This article was edited by Editorial Team. For more information on how we create and review content, see our