Watch this video to better understand Luminal graphs. The same concepts apply to LuminAIR, since LuminAIR is a compiler built on top of Luminal that adds cryptographic proofs to computation graphs

LuminAIR can prove not only ML, but any computation expressed as a compute graph. Here’s the compute graph of a simple risk assessment example proving VaR & CVaR metrics, that may help protocol governance justify parameter updates in a trustless way. github.com/gizatechxyz/Lu…

Nutritionist: “Your diet is bad.” Dentist: “Your brushing is wrong.” Friends: “Your life is a mess.” Claude: “You're absolutely right!”

What a journey 🫶 This is just the beginning

Stone and S-two -- what's the leap we're talking about? Stone is StarkWare's first production grade prover. Some numbers: - It has been running in production for over 5 years - It has settled over $1T - It saved over $1B in fees for users It's a beast. S-two is the second time

FYI - S-two is open source and there are a few teams building on it already (!). >

Nous Research partnering with Luminal. Fantastic!

Just a reminder that for two months now, Giza has been proving Rekt News's recommendation system with S-two prover (cc StarkWare (BTCFi arc)) And you can verify the proofs from your browser on Rekt. Why does it matter? We believe that no news site or social platform should bias

S-two is the fastest prover in the world. Period

New LuminAIR example: A zkML implementation that uses Physics-Informed Neural Networks to solve the Black-Scholes partial differential equation for option pricing. The proof of inference is generated via Giza LuminAIR with StarkWare (BTCFi arc) S-two prover. PINNs are neural

This is how the compute graph being proven looks like. ✅ Trace generated in 5.063625ms ✅ Proof generated in 499.265167ms ✅ Proof verified in 514.208µs Note that LuminAIR is still in development. In the next phase we will fuse operators to reduce the graph size and improve

Seeing some of the comments, I realize there might be some confusion or misunderstanding about what this means. I confirm and re-iterate that this is an incredible performance and that no other proving system in the world today can achieve such feats with this level of

SHA256 benchmarks for Cairo M are here Check the notebook, run locally, make your mind about the perf of Kakarot new zkVM for client side proving dedicated precompile coming soon github.com/kkrt-labs/zkvm…

Pendle is complex. We make it easy to use. Our new agent, Pulse, autonomously allocates your deposits across PT markets. usepulse.xyz

Fib 59452 in 5.27 s on my iPhone 13. Very cool Hyli ! Cairo M is very promising Kakarot 👏 LFG

New milestone unlocked! We've implemented a STARK verifier for StarkWare (BTCFi arc) 's Stwo prover in Simplicity by Blockstream Research This is not a toy exercise anymore – it's a real modern proving system. More about the results and challenges 🧵👇

With our zk-ML work and LuminAIR framework, we’ve built: - Trustless content recommendation systems - Provable algorithm for climate claim protocol - Intelligent AMM solvers As you look to bring verifiable intelligence into your agents, oracles, or protocols, build with us.

In case this was not clear (surprisingly): ZK and FHE >>>> TEE. Pure math >>>> manufactured hardware. Trust the math. If you want to see the integrity web becoming a reality, it will be with ZK. Prove the world, one STARK proof at a time.

Amazing milestone for Starknet (BTCFi arc) Congrats StarkWare (BTCFi arc), S-two is an amazing frontend and backend framework for generating STARK proofs

Congrats guys! Incredible work so far 👏