Spent the last few days digging into @OpenGradient , mostly because I'm tired of projects slapping "verifiable AI" on a deck without explaining what that actually means. This one made me sit up a bit.

Here's the simple version of what they're doing: when you ask an AI model a question, you're usually just trusting whatever comes back. You can't check if it actually ran the model it claims to, or if someone tweaked the output along the way. OpenGradient tries to fix that by attaching a kind of digital receipt to every single AI response proof of exactly what model ran, what you fed it, and what it spit out. You don't have to take anyone's word for it.

What caught my attention first was the funding. They've pulled in $9.5 million total, with a16z crypto and Coinbase Ventures backing them , which isn't a small signal in this space. But funding alone means nothing if nobody's using the thing.

That's actually pretty interesting because the usage numbers aren't nothing either — over 2 million verifiable inferences run so far, with 500,000+ proofs generated across more than 2,000 hosted models . The OPG token launched its TGE back in April 2026, and Binance picked it up early, which gave it real liquidity out the gate.

Most projects talk about "ecosystem growth" without showing receipts. Here, over 100 developers have contributed models, and the network apparently has six different ways it generates revenue which at least suggests people are building, not just speculating.

Whether it translates into lasting adoption remains to be seen, but at least they're shipping.

#OPG @OpenGradient $OPG