The Most Interesting OPG Number Isn't 2 Million
When people talk about OpenGradient, the number that usually gets highlighted is the 2M+ verifiable inferences processed by the network.
What caught my attention was a different number sitting right beside it.
More than 500,000 zkML proofs and TEE attestations.
Maybe it's because that figure feels closer to the actual problem OpenGradient is trying to solve. AI today works surprisingly well, but most of us still have no way to verify what happened behind the output. We trust the provider, and that's about it.
OpenGradient's approach seems to be building around verification itself. Inference nodes handle the computation, while full nodes verify the result before it gets recorded on-chain. Developers can choose between zkML proofs or TEE attestations depending on the balance they want between cost and security.
What I keep wondering about, though, is how that 500K figure is distributed.
zkML and TEE aren't really the same thing. One relies on mathematical proofs. The other relies on trusted hardware. Both have value, but they offer different trust assumptions.
That's why I'm less interested in the headline number and more interested in the composition underneath it. As the network grows, it will be interesting to see whether demand for verifiable inference grows naturally and whether fee generation can keep pace with adoption.
Either way, verifiable AI remains one of the more interesting ideas developing at the intersection of crypto and AI right now.