Anthropic releases FlashProve, an open-source zkSNARK system for provable inference
READ THE ORIGINAL SOURCEOpen original sourcewww.anthropic.com · Opens in a new tabAnthropic describes FlashProve, an open-source prototype for provable inference using zkSNARKs, which lets a model provider commit to a model and prove responses came from it without revealing weights. For Llama 4 Maverick (400B-parameter mixture of experts, 4-bit quantized), a single H100 generates proofs at about 110 tokens/second, with a proof size of roughly 19 MB per 1,024-token request. The release includes a prover, two verifiers (C++ and Lean), Lean proofs of top-level security statements, and a protocol description under Apache 2.0.
One primary source; other links are context, not independent corroboration.
- Primary sourceX discovery · automated brief
- Discovery linkX
- Linked in sourceprotocol description
- Linked in sourceExecutable code
- Linked in sourceprover
- Linked in sourcefast verifier
- Linked in sourceprovably correct verifier
- Linked in sourceLean proofs
- Linked in sourceinteractive visualizer
- Linked in sourcetrust challenges
- Linked in sourceregulate
- Linked in sourcepace
- Linked in sourcemonitoring hardware and how it is used
- Linked in sourcedevote its hardware to inference
Grok Bot submitted the link. AINewsIQ fetched the primary source and prepared a source-based brief automatically. An editor has not reviewed the story; social attention and the brief do not independently verify its claims.
Open full article view →- Linked article+30
- Fetched source page+30
- No editor link review+0
- Original non-discussion link+5
This measures the stored source trail and editorial steps. It does not certify accuracy, peer review, or independent corroboration. Popularity adds no points.
View full scoring →Read the scoring rubric →What the source says
Anthropic describes FlashProve, an open-source prototype for provable inference using zkSNARKs, which lets a model provider commit to a model and prove responses came from it without revealing weights. For Llama 4 Maverick (400B-parameter mixture of experts, 4-bit quantized), a single H100 generates proofs at about 110 tokens/second, with a proof size of roughly 19 MB per 1,024-token request. The release includes a prover, two verifiers (C++ and Lean), Lean proofs of top-level security statements, and a protocol description under Apache 2.0.
A READER'S EVIDENCE CHECKHow to read this signalSix questions about claims, sources, risks, and open checks
This is a signal-level guide until claim-level evidence is available. “Not recorded” means the evidence has not been assessed here; it does not mean a result is false. How we check AI news →
- 01Who is making each claim?
The claimant and whether each claim was measured or asserted are not recorded in this public brief.
- 02What was actually tested?
Not recorded per claim. Check the source for sample size, tasks, model and benchmark versions, and reproducibility.
- 03Where are the sources?
Original link: www.anthropic.com. Brief basis: fetched source page. Primary page date and quote locations are not established in this panel.
- 04What could mislead?
Source claims not independently checked; the reported performance figures come from the authors' own benchmarks.
- 05What is at risk if wrong?
Not assessed per claim. Consider safety, privacy, cost, and adoption decisions before relying on the result.
- 06What would change the view?
No claim-specific check is recorded. An inspectable primary method or independent replication could change the assessment.
Key findings from the stored brief
- A single H100 GPU generates proofs at around 110 tokens/second for Llama 4 Maverick in 4-bit quantized weights.
- Proof size is about 19 MB per 1,024-token request, per the source.
- Proof generation costs roughly 70-95x the GPU time of running the model on vLLM, according to the source.
- Five artifacts are released: a blog post, technical report, protocol description, executable code with two verifiers, and Lean proofs.
Source claims not independently checked; the reported performance figures come from the authors' own benchmarks.
This is a source-based abstract, not independent validation of the source's claims.
Evidence & sources
These links show where the signal came from and what the source page linked to. A linked page is context, not independent corroboration.
Related archive coverage
These relationships record a citation or topic overlap, not confirmation of the finding.
Grok Bot submitted the link. AINewsIQ fetched the primary source and prepared a source-based brief automatically. An editor has not reviewed the story; social attention and the brief do not independently verify its claims.
These are Grok Bot discovery notes. AINewsIQ prepared a brief from available source material. The bot's rationale, linked posts, authors, and stances have not been independently verified.
Anthropic Frontier Red Team post introduces FlashProve, reporting ~110 tok/s proofs on Llama 4 Maverick (400B MoE) on one H100 with ~19 MB proofs. Steve Weis post drew 393 likes; HN item 50348831. Page dated Nov 2, 2026 (JSON-LD midnight).
Source claims not independently checked; the reported performance figures come from the authors' own benchmarks.
Timeline & updates
- Signal dated by source or discovery feed
This timestamp places the signal in the radar. It may be the discovery post date rather than the primary page's first publication date.
- Shared brief stored
AINewsIQ prepared one abstract from fetched source page.
A publication revision history is not stored for this signal. New engagement snapshots do not mean the source itself changed.
Why this may matter
Verifiable inference could let users, evaluators, and regulators check that a claimed model actually served a response, supporting trust, compute-monitoring for pacing proposals, and tamper detection. The source frames this as intermediate progress shared for others to build on.
Limits and open questions
Source claims not independently checked; the reported performance figures come from the authors' own benchmarks.
Organization-specific financial, security, and operational effects require the organization's own context and review.
Discussion
Comments are reviewed before publication. Discuss the evidence, not the person.
No published comments yet.
Reader discussion opens with accounts. You can read the original source and published comments now.