AIRankBankEvery launch is a claim. Every claim gets a verdict.

In a post published in November 2026, Anthropic states that FlashProve generates proofs at around 110 tokens per second on a single H100 GPU for Llama 4 Maverick, a 400-billion-parameter mixture-of-experts model, with a proof size of about 19 MB per 1,024-token request.

✓ DELIVERED · settled 2026-11-12

Claim made inAnthropic releases FlashProve, an open-source prototype for proving which model answered (anthropic.com)
Bank recordAnthropic — see its full delivery record in the bank
Market opened2026-11-04
Market typeWeekly market (7 days)
Calls lock2026-11-10 07:00 Eastern Time (12:00 UTC) — 24 hours before the check, so nobody can call it with the answer already visible
Check time2026-11-11 07:00 Eastern Time (12:00 UTC)
Source of recordanthropic.com

Background

Here is an answer to a question users could never really ask: how do you know the model that replied is the model you paid for? Anthropic's FlashProve attaches a cryptographic proof to each response, a zkSNARK that ties the output to a published fingerprint of the model's weights without revealing the weights themselves. For a 400-billion-parameter model, one H100 generates proofs at around 110 tokens a second, each proof about 19 MB per 1,024 tokens — roughly 70 to 95 times the GPU cost of just running the model. That is far too expensive to ship tomorrow, and Anthropic says so itself. But as a tool for auditors, evaluators and regulators who need to know which model really ran, it turns a trust-me claim into something checkable, and the code is Apache-licensed for anyone to pick apart.

This market exists because launch announcements are easy to make and hard to grade. Press releases describe what a company intends to ship; they rarely come back later to say whether it shipped. AI Rank Bank pulls one checkable sentence out of the coverage — the sentence above — gives it a date, and lets readers price it in public. The launch itself stays with its publisher; what lives here is the claim, the crowd's call, and eventually the receipt.

How this claim is judged

Judging criteria, fixed when the market opened: Delivered if, at the check time, Anthropic's post remains published stating the around-110-tokens-per-second proof rate on a single H100 and the about-19-MB proof size per 1,024-token request.

The source of record is the publisher's own page linked above. A claim about a company's product is judged against that company's public output — its product pages, documentation, changelogs and official posts — not against rumors or third-party summaries. Only yes-or-no outcomes count: partial credit is not a verdict, and a claim that needs a paragraph of excuses has already answered itself.

How readers called it

1 reader called this claim. 100% said it would deliver.

1 Will deliver · 0 Won't ship · final split when the market locked

Swift-Falcon-98Will deliver50% agreed · 2026-11-05✓ +50 pts

Points reward being right against the crowd. A correct call pays 100 × (1 − the share of readers who agreed with you when you placed it), with a floor of 10: side with a 90% consensus and a win pays 10, stand alone and a win pays up to 100. Wrong calls pay nothing. The price is locked in when the call is placed, so early readers are paid for judging before the crowd forms, not for copying it.

How readers split

Share of readers saying “will deliver”: opened at 100%, now at 100%. 86 snapshots on record since 2026-11-05 — a snapshot lands whenever a call is placed, and at least hourly while the market is open.

Verdict

✓ DELIVERED — checked against the source of record and settled on 2026-11-12. The evidence is public: https://www.anthropic.com/research/provable-inference. Readers placed 1 call on this claim; 100% of them said it would deliver. The market paid out 50 points in total. Verdicts are permanent. If the evidence ever shows this stamp was wrong, the record is corrected openly and the correction stays on this page.

About this market

AI Rank Bank is an AI news front page with a memory. Every launch on the homepage carries one claim like this one, with a check date. Readers call claims with points, build a public record, and climb the leaderboard. Companies build a record too: every settled claim lands on the company's bank page, where its delivery rate is simple arithmetic that nobody gets to edit afterwards. There is no money in the market — points buy bragging rights only, which keeps the incentives pointed at being right.