中文

← All funding rounds

Lanyon AI

United StatesScience models and AI scientistsPrinceton, New Jersey, United StatesPrivate

Website · lanyon.ai ↗

A research lab building an agent that writes scientific solvers and their formal proofs together, so the code provably matches its specification; aimed first at aerospace, propulsion and nuclear simulation.

Disclosed total$10.6M
Rounds on record1
Latest round$10.6MRound (unnamed) · 2026-08-17
Last stated valuationNone stated
Investors named3
Research on record1highest level 2 · Computed

Money and evidence

Moneyeach round in US dollars; ≈ converted at its date$10.6MEvidencelevel of each research entry12345
Closed roundResearch entry, at its level

Latest round, Round (unnamed), 2026-08-17: the record then held 1 research entry, highest level 2. The first research entry is dated 2026-08-03; the highest level so far is 2 (Computed).

Rounds and research, on one timeline

  1. 2026-08-17Round (unnamed)
    Round

    $10.6M

    For: not itemised; the release describes it as the initial round on emerging from stealth

    Not stated

    Amount, date and lead are from the company's release of 17 August 2026, which names Dimension and Industrious Ventures; the angel Siqi Chen is named only in the company's blog post on its investors, which dates the close to July 2026. No valuation is given.

    Lead · DimensionIndustrious VenturesSiqi Chen

    On the deal: Simon Barnett (Partner and Head of Research; spoke for the firm in the announcement, Dimension); Siqi Chen (Angel investor); Zavain Dar (Partner, Dimension)

    By this date the record held 1 research entry, highest level 2

    Announced by the company · prnewswire.com ↗On the timeline →

  2. 2026-08-03
    Research

    Level 2 · Computed

    Lanyon AI's own benchmark has five frontier models write PDE solvers with Lean proofs, and finds they often prove something other than the code they wrote

    The test targets a failure this record tracks: a proof that type-checks but is about different code. Under the terse prompt, both Claude models proved a first-order, CFL=1 special case for the second-order solver they had written, and some GPT and Kimi runs used native_decide to skip kernel checking. It shows what comparing a proof against the code can catch, and how much the prompt decides.

    Timeline entry →lanyon.ai ↗

Research entries are the timeline's graded entries that name this company. A level grades one claim, not the company, and a blank can mean the record has not caught up rather than that nothing was published.

See these rounds in the table →

Its investors also backed

Other companies on this page whose rounds name the same investor.