01 / The mechanism and its boundary
What the technique establishes
Pearl is a blockchain developed by Pearl Research Labs whose mining is designed to be a by-product of the matrix multiplications that AI models perform on GPUs. Each block carries evidence that a certain amount of matrix-multiplication work was done. Its network is live and its code is public. Pearl reports that a mining variant of a 70-billion-parameter model served more tokens per second on four H200 GPUs than the original model, in a benchmark that does not isolate the cost of mining. An independent study found that the network's dominant mining software appears to multiply random matrices, which verification does not distinguish from AI work. Its R3 (in production) rating is for proving that matrix-multiplication work was done on a blockchain, not for verifying how AI hardware is used. Its security rests on new hardness assumptions, and Pearl lists known shortcuts that let miners do less work than nominal.
- Threat model
- Adversarial prover
- Adversarial evaluation
- Published analysis
- Hardware needed
- None
- Prover cooperation
- Required
- Confidentiality
- Partial
- Category
- Cryptography & computation
Claims and scope
A direct link identifies the intended claim. A supporting link supplies part of the evidence. Neither establishes that a complete verification system has been demonstrated.
This compute runs inference, not training
Demonstrates matrix-multiplication work proofs, a component proposed for capacity accounting (S-1102); Pearl has not demonstrated that use.
Readiness for a stated use
Assessed use: checking matrix-multiplication work proofs for blockchain consensus
low confidence · current · assessed 2026-10-08 · rubric 1.1
This is the source map’s editorial assessment. Production use is not evidence of resistance to every adversary.
R3 for checking claims of matrix-multiplication work in blockchain consensus. The network is live, but no independent security evaluation exists, and the capacity-bounding use in Proofs of useful work for capacity accounting is not demonstrated.
- R1 met: the protocol, verifier and hardness assumption are specified S-1105, building on a published construction S-1609.
- R2 met: code for a full node, a GPU miner and a proof-of-work circuit and verifier is public S-1107. Pearl reports running it on H200 GPUs alongside LLM serving S-1106, against miners who try to win more often than honest work allows S-1105.
- R3 met for the narrow claim, as production software that is available and relied on by other parties. Pearl reports that the chain went live when the node code became public S-1106 S-1107. The repository supports mainnet, and no release up to v1.2.1 is labelled alpha, beta or preview S-1107. An independent study counted 8,012 online workers among the top 15 miners of one mining pool in May 2026, ran experiments on mainnet, and had shares accepted by that pool, whose acceptance rests on the protocol's verification S-0071.
- R4 not met: the README mentions no audit S-1107, and that study measures how the network is used rather than auditing or attacking the protocol S-0071. As of September 2026 no independent security evaluation has been published.
Evidence needed for the next level
An independent public security evaluation of the protocol or its implementation.
For verification use: a demonstration that proof rates can bound the spare capacity of declared hardware.
Limitations, flaws, and blockers
These are attributed assessments from the source map. Absence of a listed flaw is not a security guarantee.
minor / open / theoretical argument
Known mining speedups reduce work per proof
Pearl lists known speedups: crafted inputs, precision shortcuts, seed or commitment grinding, work reuse, and faster kernels or hardware. Its jackpot policy checks limit crafted inputs, and a policy check caps skippable summands at one-sixteenth of those in a tile S-1105. Pearl describes faster honest kernels or hardware as "not an attack on the protocol" S-1105.
significant / open / open question
Security rests on a new, informal hardness assumption
The FP8 scheme relies on "Assumption 1 (Informal quantized-subspace hardness)": quantised products of noised matrices are assumed not to be substantially easier than generic ones S-1105. The integer construction it extends lists PoUW from more standard assumptions as an open problem S-1609.
minor / open / theoretical argument
Verification does not check that mined matrices come from AI workloads
Miners choose their own matrices S-1609. Basu reports that Pearl's verification "does not check whether the matrices originate from an AI model", that random matrices pass it, and that Pearl's reference mining code generates uniformly random matrices, with vLLM inference as an option. String analysis suggests that the dominant third-party mining software contains no inference code. Basu also finds that a naive fixed-threshold check of matrix kurtosis is defeated, at negligible cost, by sampling clipped Gaussian matrices. Basu calls the gap "a design property" rather than a vulnerability S-0071. It does not affect the claim that work was performed, but it means the "useful" part of the work is not verified.
What still blocks use or stronger assurance
- S-1102
Built for consensus rather than capacity bounding; verifying that declared hardware has no spare capacity would also need a credible compute estimate.
- S-1106
Performance figures are provider-reported, and the benchmark reports no baseline of the certified model without mining.
- S-1105
Bit-exact verification depends on reproducing GPU arithmetic deterministically.
Connections in the research map
Depends on
- Deterministic and bit-exact inference
The verifier recomputes a tile bit for bit, reproducing GPU arithmetic off the GPU with the Hawkeye technique.
Mechanisms implemented
Organizations and developers
Sources and provenance
- S-1105 / Tier B
Pearl Floating Point Scheme Specification ↗
Pearl Research Team · 2026 · Pearl Research Labs
Supports: FP8 protocol, verifier, jackpot policy, known speedups, ZK verifier, assumption (provider-reported)
Locator: abstract; §1; §2; §4.1; §5.1; §6; §7; App. A–B
Version and catalogue details - S-1106 / Tier B
Pearl INT Whitepaper ↗
Pearl Research Labs · 2026 · Pearl Research Labs
Supports: INT scheme, launch statement, benchmark table, zkSNARK size (provider-reported)
Locator: §3; §4.2; §4.7; §7 Table 1
Version and catalogue details - S-3571 / Tier C
Proof of Useful Work from the Ground Up ↗
Pearl Research Labs · 2026 · Pearl Research Labs
Supports: provider-reported end-to-end serving overhead for the integer proof-of-useful-work kernel
Locator: Measured in production
Version and catalogue details - S-1107 / Tier B
pearl: Monorepo for the Pearl network ↗
Pearl Research Labs · 2026 · GitHub
Supports: public node, miner and ZK prover code; releases
Locator: README; release v1.2.1 (commit e72ad36)
Version and catalogue details - S-1609 / Tier B
Proofs of Useful Work from Arbitrary Matrix Multiplication ↗
I. Komargodski, O. Weinstein · 2025 · arXiv
Supports: underlying PoUW construction and its assumptions; miners choose the instance
Locator: abstract; §2; §3
Version and catalogue details - S-0071 / Tier B
The Usefulness Gap in Proof-of-Useful-Work: An Empirical Study of Pearl's cuPOW Protocol ↗
A. Basu · 2026 · arXiv
Supports: independent measurement of Pearl's mainnet (integer scheme): worker count, mining software, random-matrix mining, verification scope, statistical check defeated
Locator: abstract; measurement and verification sections; conclusion (v2)
Version and catalogue details - S-1010 / Tier A
Hawkeye: Reproducing GPU-Level Non-Determinism ↗
E. Badash, D. Boneh, I. Komargodski, M. Srivastava · 2026 · Proceedings of Machine Learning and Systems 8 (MLSys 2026)
Supports: Hawkeye, the bit-exact GPU reproduction the FP8 verifier uses; Komargodski's affiliation with Pearl Research Labs
Locator: author block; abstract
Version and catalogue details - S-1102 / Tier C
Pacing AI Requires Proof ↗
Attestable · 2026 · Attestable blog
Supports: a proof of some computation is not a proof of all computation; capacity bounding needs a compute estimate (Attestable's proposal)
Version and catalogue details
- Source review date
- 2026-09-25
- Drafted by (source map)
- ai
- Review handles (source map)
- codex-review