Brevis’ Pico Prism Achieves Real-Time Ethereum Verification with Massive Cost Reduction
QuickBreakdown
- The Brevis Network announced that its Pico Prism zero-knowledge Virtual Machine (zkVM) has achieved a 99.6% proving coverage.
- This breakthrough performance reduces the necessary GPU hardware cost for real-time proving by 50%
- Pico Prism’s advancement moves Ethereum’s verification model from re-execution to a more efficient, single-time verification.
Brevis, a verifiable computing platform, has announced a significant breakthrough with its Pico Prism zero-knowledge Virtual Machine (zkVM), achieving near-total real-time proving coverage for current Ethereum mainnet blocks. The company stated that Pico Prism achieved 99.6% proving coverage in under 12 seconds and 96.8% coverage in under 10 seconds across 45 million gas blocks. This performance leap is a key step toward realizing the goal of a Layer 1 zero-knowledge Ethereum Virtual Machine (zkEVM).
A few weeks ago we revealed the ProverNet whitepaper for the first time.
Today, Brevis ProverNet mainnet beta is live!
Our decentralized marketplace for ZK proof generation is now open to provers & applications🧵 pic.twitter.com/58PmwAUsvb
— Brevis (@brevis_zk) December 8, 2025
The performance improvement substantially reduces the economic barrier to entry, eliminating the need for expensive, high-end hardware. The advancement reduces the GPU hardware costs by 50%, making real-time processing more economically viable and accessible for widespread deployment, even on consumer-grade hardware. This shift is critical because it supports Ethereum’s transition to a system in which all verification is handled through cryptographic proofs.
Cost reduction and scalability enhancement
The Pico Prism’s success stems from a complete architectural redesign that transitions from single-machine proving to distributed multi-GPU clusters. By breaking the proof process into parallel tasks, the modular design leverages the power of multi-GPU clusters, shifting computation-heavy workloads to Graphics Processing Units (GPUs). This modular approach is designed to adapt to future needs, combining a general-purpose zkVM with a data coprocessor to handle various application scenarios.
By enabling a single prover to generate a mathematical proof that all nodes can verify in milliseconds, Brevis fundamentally changes the validation process. This replaces the current model, where hundreds of thousands of validators must re-execute identical transactions, significantly increasing transaction throughput and lowering the barrier for home stakers. This innovation is already being utilized by applications such as PancakeSwap’s advanced trading hooks and Frax’s cross-chain verification systems.
Meanwhile, Starknet’s v0.14.0 upgrade, after community approval, enhanced the decentralisation, speed, and performance of the Ethereum Layer 2. Key features included adopting Tendermint for decentralised sequencing for better fault tolerance and censorship resistance, introducing subsecond pre-confirmations, and implementing an EIP-1559 fee market to stabilise gas prices by decoupling them from Ethereum’s volatility. This release advanced Starknet towards complete decentralization of the sequencer as a top scaling solution.
Take control of your crypto portfolio with MARKETS PRO, DeFi Planet’s suite of analytics tools.”
Disclaimer: The content of this article solely reflects the author's opinion and does not represent the platform in any capacity. This article is not intended to serve as a reference for making investment decisions.
You may also like
Has the poor performance of new stocks in the same track dampened subscription enthusiasm? Bamboo makes an emergency halt the night before pricing, doubling its valuation in a year but struggling to find a "buyer".
Bamboo, the master underwriter for homeowners insurance in California and Texas, postponed its IPO on Tuesday. The company had planned to issue 35 million existing shares at $18–$20 per share to raise $665 million.

Is the AI replacement wave overestimated? Barclays: Only 20% of core workplace skills can be highly replicated, but amplification effects are quietly spreading
The impact of AI on the workforce is milder, yet more profound than previously imagined. Barclays' latest research dispels the panic of a "replacement wave"—only about 20% of core professional skills can truly be replicated by AI, but its "amplification effect" permeates almost every corner of every profession. What is more concerning is that the demand for AI skills is spreading from tech positions to management roles, and the rise of physical AI could even make human professional skills a new, scalable form of intellectual property.
Costco Q4 Earnings Preview: Membership and Capital-Return Test After a Near-Bear Pullback

