I went downstairs to a convenience store and bought oden, came back upstairs without even taking off my shoes—the computer screen was still on, showing the developer panel for @OpenGradient . I put the soup in the corner of the table and sat back down.
Originally I just wanted to glance at the number of new projects connected today, but I paused on the line that said 100+ developers. I exported the data from the past three weeks and compared it side by side, five times over. The masses stare at 100+ developers and use that number as a signal that the ecosystem is growing—like back in the day when you’d have an instinctive reaction to TVL. I’m not looking at this frontend metric.
What I really want to confirm is this: every time these developers call a model, does the underlying Verification Layer run as well? In other words, the number of models on the Model Hub rising from a few thousand to 4,500 isn’t that meaningful; what’s truly scarce is that every single inference is accompanied by a TEE attestation or a zkML proof landing on-chain. The two figures disclosed by OpenGradient—two million verifiable inferences, and five hundred thousand zkML proofs with TEE witnesses—the ratio between those two numbers is the real proof of work for this network.
I pulled the developer growth curve and the proof generation curve onto the same chart and compared them in three rounds—the latter’s slope didn’t fall behind. New developers aren’t just playing around with a few tweaks in the frontend SDK; their requests really run through the complete path from inference to witness to on-chain submission. That path is the underlying backbone that makes OpenGradient different from ordinary AI API platforms.
If $OPG is only responsible for node incentives and gas offsets, then it’s more like a general network fuel token. But if in the future the model hosting fees, inference verification settlement, agent-call pricing, and data-access authorization all form a closed loop around it, then it’s no longer just fuel—it becomes the settlement unit for the entire verifiable AI network.
The oden is already half cold. No rush to draw a conclusion. Developer numbers will keep rising, and the narrative will change too. I’d rather wait for the mainnet to reliably hand over proofs for every single inference, then decide how to look at all of this. #opg