Here we report on the progress of the leading builders in the Proving Service ecosystem, documenting recent significant releases, technical breakthroughs and general updates
@fermah_xyz has integrated Jolt, a zkVM developed by @a16zcrypto, to streamline ZKP generation, making it faster and more affordable. Generating ZKPs has traditionally been complex and costly, but Fermah simplifies the process by managing the entire proving workflow: https://fermah.xyz/blog-posts/jolt-x-fermah…
They also announced an integration with @Scroll_ZKP to enhance the network's proof pipeline, ensuring efficient validation of transactions. Fermah’s intelligent orchestration optimizes hardware use and enables seamless proof generation at scale: https://fermah.xyz/blog-posts/fermah-powers-scroll…
Finally, the second episode of 'From Zero to Knowledge' was released. @vanishree_rao, Co-founder of @fermah_xyz and seasoned cryptographer, shares insights into ZK’s evolution from a theoretical concept to a core component of modern digital infrastructure, and discusses its future trajectory.
This educational series, supported by @HouseofZK, is designed to make ZK principles clear and accessible to developers of all backgrounds, helping newcomers build a strong foundation in the field: https://hozk.io/education
@thezkcloud has announced the 'Request for Comments' for ZkBoost, an open-source unified client for outsourced ZKP generation: https://blog.zkcloud.com/p/request-for-comments-zkboost-unified…
The initiative aims to standardize interactions with proving services, reducing fragmentation and increasing healthy competition. By abstracting integration complexities, ZkBoost ensures better pricing for users.
@Niall_Emmart, Co-founder of @Snarkify_ZKP, shared a deeply insightful thread discussing GPU speed limits in proving systems: https://x.com/Niall_Emmart/status/1888925436723552363…
NIall highlighted that teams using the same hardware and algorithms achieve vastly different speedups, from 10x to 900x. Analyzing Snarkify’s @z_prize 2023 submission, he estimated a theoretical lower bound of 486 ms, with their runtime just 50 ms above it. He also emphasized the importance of hardware-aware optimization for real-time @ethereum block proving.
@cysic_xyz has partnered with @P2Pvalidator for enhanced ZKP generation: https://medium.com/@cysic/strengthening-decentralized-proofs-p2p-joins-cysic-network-40549e4c5976…
P2P’s institutional-grade expertise aims to help strengthen Cysic’s scalability, decentralization, and efficiency, making ZKPs more accessible and cost-effective.
@SindriLabs introduced zkVM-as-an-API, enabling quick zkVM deployment. Developers can build with @SuccinctLabs SP1 and @a16zcrypto Jolt through a unified interface featuring serverless execution, SDK and CLI access, and built-in CI/CD: https://x.com/SindriLabs/status/1892227395068772820…
Additionally, Sindri API 1.17.0 now supports compiled Leo programs, allowing developers to use SnarkVM, @AleoHQ’s zkVM. Leo, a Rust-like language developed by Aleo, simplifies the development of ZK applications: https://sindri.app/changelog/2025/02/17/snarkvm-support/
@Ingo_zk has released ICICLE V3.5, featuring a CUDA-optimized Sumcheck API for functions over multilinear polynomials. This update enhances ZKP performance and introduces Proof-of-Work for FRI-like protocols, the Poseidon2 sponge function, and various bug fixes. It also marks the first use of Fiat-Shamir in ICICLE, now under audit. Future updates will add Rust, Go, and Metal backend support for Apple Silicon: https://ingonyama.com/blog/icicle-v3-5-sumcheck-with-lambda-functions…
They also introduced AIR-ICICLE, integrating the Plonky3 framework with the ICICLE field library for efficient trace generation and symbolic constraints. This enables users to write AIR circuits in Plonky3 while leveraging ICICLE’s GPU acceleration for STARK proving. The integration simplifies execution trace compatibility, eliminating costly data conversions. Future work includes a backend prover and deeper ICICLE-Plonky3 synergy for optimized ZKPs: https://hackmd.io/@Ingonyama/air-icicle
@jimpo_potamus, Co-founder of @IrreducibleHW, appeared on the 'Into the Bytecode' podcast, discussing the development of Binius, a proof system using binary fields for verifiable computing. He explored the impact of cryptographic proofs on blockchain, hardware acceleration with FPGAs, and the broader vision of a verifiable internet: https://youtube.com/watch?v=_Rqp7tanSwg
@MarlinProtocol now powers @UngateAI's Wukong launch, enhancing AI agent verification through its Trusted Execution Environment network: https://x.com/MarlinProtocol/status/1892555145780580663…
By generating cryptographic proofs and validating them via @eigenlayer operators, this integration ensures agent autonomy and execution integrity. With three DeFAI projects already live, Wukong is seeing strong adoption in capital formation, governance, and personal finance apps.
@zan_team showed a 15x speedup for @Starknet's Stwo prover using their GPU network. Benchmarks show 3090 and 4090 GPUs vastly outperform CPUs, reducing computation time significantly:
