❄️ ICICLE v4 is here, and it’s our biggest release yet. ➡️ https://t.co/NeUNZc3tFh

Result: 16 MiB blob encoded in ~1.26 seconds using @Ingo_zk's ICICLE library. For scale: a naive CPU-only encoding of a 128 KiB blob takes 8.2 seconds 56× slower per-byte at 1/128th the size. GPU acceleration is our path to a true production system.

Data availability hits a wall: validators need to verify data is available, but storing full copies across hundreds of nodes doesn't scale. EigenDA solves this with Reed-Solomon + KZG: validators store fragments, but anyone can verify authenticity and reconstruct from subsets.

EigenDA needs to scale securely, the core primitives to do so are Reed-Solomon Encoding & KZG Commitments. In order to implement this needed to turn theory into implementation with GPU Acceleration leveraging @Ingo_zk's ICICLE

“The infrastructure is ready!” Congrats team on this milestone ⚡️

Today @eigen_da reaches cloud scale at 1GB/s. Benchmarked across 100 operators and 10 countries. ft. @sreeramkannan and @0xkydo

🚀 Excited to be at ZKARCH, the 1st Workshop on Zero-Knowledge Proofs & Verifiable Computation! 📍 Seoul, Korea 📅 Oct 18, 2025 🌐 In conjunction with MICRO 2025 🎤 Speaker: @shanie_winitz 📝 Talk: Accelerating the Sumcheck Protocol in Hardware https://t.co/Y4TFhg8u3A

Needed something to read on sumcheck? See the latest posts covering some basics on multilinear plus this helper to understand @Ingo_zk 's paper with optimizations.

Least Authority CEO @liz315 and Security Researcher @kaplannie joined the SNARK Chocolate podcast from @Ingo_zk to talk about our roots, pathways into security auditing, privacy-first audits, cryptography trends, and the limits of AI in security.

Here’s our talk from Google ZK Summit last week: High-Speed Cryptography / @OmerShlomovits https://t.co/jsqNhf677K H/t Mauro for the recording, Google for hosting.

📢Announcing liboqs x ICICLE ML-KEM integration liboqs is an open-source C library for quantum-safe cryptographic algorithms, part of the Linux Foundation's PQC Alliance! Links: 🌻ICICLE ML-KEM: https://t.co/jXacFJLsM4 🌻liboqs: https://t.co/P7XO0RQ1rD 🌻Integration PR:

exactly one year ago I joined @Ingo_zk. in that time, i have worked on many projects. but two are special: icicle-snark and icicle-halo2. when i first learned about zk i was trying to understand what is r1cs, how groth16 works, how to write circom circuits. today i'm author

Catch @ravitals on the Snark Chocolate podcast talking about the state of Fully Homomorphic Encryption (FHE) adoption 👇

🍫 Season 1 of Snark Chocolate is complete. Thanks to all our guests and listeners for making it a success. 🎧 Listen now: 🟢 Spotify: https://t.co/lAZAAszzNK 🟣 Apple: https://t.co/g0WRjR38QB 🔴 YouTube: https://t.co/nbnNUbOkJI 💬 Continue the conversation: Our exclusive,

🍫 New Snark Chocolate Episode 🍫 Sunscreen’s FHE Future This week we sat down with @ravitals, founder of @SunscreenTech, to talk about the road to real-world FHE adoption. 🔐 Past, present & future of FHE 🏗️ Sunscreen’s architecture & design 🛠️ Tech stack & roadmap ✨ What

.@Ingo_zk presented ICICLE-Halo2 v2, a new fork of ezkl-halo2 with up to a 25x performance boost. Based on the popular Halo2 zk-SNARK originally by ZCash, this version brings GPU-accelerated improvements applicable to other versions like zcash-halo2, pse-halo2, and axiom-halo2.

@fermah_xyz @zksync @brevis_zk @QuickswapDEX @soneium @cysic_xyz @SuccinctLabs @zan_team @AnthropicAI @thezkcloud @ethereum @vlayer_xyz @zerobasezk @alignedlayer @eigenlayer Ingonyama 🔹 Tech 🔹 @Ingo_zk has introduced a major update to its ezkl-halo2 fork, achieving up to 25x faster proving performance by making Halo2 nearly fully GPU-native: https://t.co/huSKpTd0FC Using ICICLE and custom CUDA kernels, the team offloaded key bottlenecks like

1/🚀 GPU-Accelerated ZK Proofs on Mobile is here! Check out our deep dive into Metal MSM v2: https://t.co/79N0kRKnLA Benchmarks on an M3 MacBook Air with 24GB memory show 40x–100x improvements over v1 and 10x improvement over equivalent MSM implementation in WebGPU by ICME Labs

icicle + halo2 = 25x speedup

📢 Introducing ICICLE-Halo2 v2 Our almost-native GPU, blazing-fast Halo2-EZKL Fork. 🏎️Technical writeup: https://t.co/4ptpmBctkU 👩‍💻Code: https://t.co/AmVELb0yd8 🧑‍🍳Chef: @0xemirsoyturk | huge thanks to @ezklxyz !

🍫 New Snark Chocolate Episode 🍫 Tinfoil's Confidential AI This week we sat down with Jules Drean, co-founder of @TinfoilAI, to talk about building privacy-first AI infrastructure. 🔒 Private inference via secure enclaves + attestation 🧠 Tinfoil vs. FHE vs. MPC: performance