Brevis’ Pico Prism Achieves Record 99.6% zk-Proving Coverage for Ethereum Blocks

• Pico Prism zkVM is the first in the world to achieve 99.6% proving coverage (sub-12 seconds) and 96.8% real-time proving coverage (sub-10 seconds).
• Due to the 50% reduction in GPU hardware prices brought about by this performance improvement, real-time is now economically feasible for broad deployment.
• A comprehensive architecture overhaul that switches from single-machine proving to distributed multi-GPU clusters is what gives Pico Prism its performance gains.

For current-generation Ethereum mainnet blocks with a 45M gas limit, Brevis, a leading infrastructure company that powers smart, verifiable applications with zero-knowledge proofs, announced that its Pico Prism zkVM is the first in the world to achieve 99.6% proving coverage (sub-12 seconds) and 96.8% real-time proving coverage (sub-10 seconds).

In order to meet Ethereum’s future capacity requirements, Pico Prism has advanced from research into production-ready infrastructure that removes a significant barrier in the network’s shift to base layer zero-knowledge verification. Due to the 50% reduction in GPU hardware prices brought about by this performance improvement, real-time is now economically feasible for broad deployment.

When compared to current solutions, Pico Prism offers significant gains in all important metrics:

• Coverage for existing 45M gas blocks is 99.6% (<12s) and 96.8% (<10s), whereas rival solutions tested 36M gas blocks and found coverage of 40.9%.
• Hardware costs $128K vs $256K.
• The average proving time for 45M gas blocks was 6.9 seconds; for 36M gas blocks, it was 6.04 seconds as opposed to 10.3 seconds.
• For similar performance, compare 64 RTX 5090 GPUs with 160 RTX 4090 GPUs.
• When speed and cost efficiency measurements are combined, performance is 3.4 times better.

A comprehensive architecture overhaul that switches from single-machine proving to distributed multi-GPU clusters is what gives Pico Prism its performance gains. By using Pico’s modular architecture, the system divides the proving process into parallel stages, shifting computation-intensive jobs to GPUs while keeping setup operations on CPUs.

A Leap for Ethereum

These days, when a user exchanges tokens on Uniswap, more than 800,000 validators globally independently rerun the transaction. They are all doing the same computations, reading the same states, and coming to the same conclusion. As Ethereum expands, the computational waste it causes only becomes worse. This results in Ethereum’s primary scaling limitation: block capacity remains unnaturally low as validators need affordable hardware to meet re-execution requests.

By using cryptographic proofs instead of computing brute force, Pico Prism shows that blockchains can scale verification. One prover creates mathematical proof, and everyone else validates it in milliseconds, eliminating the need for hundreds of thousands of validators to repeat the same transactions.

The ability of zkEVM to validate Ethereum blocks in real-time is crucial to the scalability of Ethereum in the future. Ethereum intends to ultimately switch to ZK proofs at every stage of the stack, from onchain privacy with client side proving to consensus layer signature aggregation. L1 zkEVM integration is the initial step in that approach.

To maintain the greatest levels of liveness and censorship resistance, the Ethereum Foundation set explicit targets in its July 2025 roadmap: 99% coverage, sub-10 second proving, hardware under $100K, and power usage < 10kW for home proving.

With Pico Prism, the route to Ethereum L1 zkEVM integration is now much more obvious. Brevis is trying to implement additional features to surpass the amended <10s real-time proving objective, even though it is still just 2.2% percentage points away.

Unconstrained Verification

Constraints are much reduced when Pico Prism handles verification:

• It is safe to raise gas limitations over the existing 45M per block.
• Transaction costs on average decline dramatically.
• Economic viability of complex DeFi operations.
• Smart contracts can access historical data without prohibitive gas costs.
• It is possible to implement privacy-preserving measures without going over user budgets.

Regarding network health and validators:

• Instead of using gaming rigs, home stakers may validate using simple laptops.
• As obstacles to participation fall, network decentralization rises.
• Without centralizing hardware, transaction throughput rises dramatically.

Now, developers may create dApps with the same degree of confidence as Ethereum L1 by using an endless supply of off-chain computing resources. What Ethereum could look like in the near future when computational limitations are removed is already being experienced by major protocols operating on Brevis infrastructure. The same proving technology powers Frax’s cross-chain verification systems, Usual’s trustless reward distribution, and PancakeSwap’s sophisticated trading hooks.

Existing smart contract blockchains can now handle infinite processing capacity thanks to Brevis, an effective and verifiable off-chain computation engine. Brevis uses zero-knowledge proofs to transfer expensive, data-intensive computations from on-chain contexts to a much cheaper off-chain engine. This allows Web3 applications to grow without interruption while maintaining the security of L1 trust assumptions.