Layer1Labs is raising seed funding — $1–1.5M on a SAFE, seeking a lead investor. Talk to us →
Guide · ChronoHive

Evaluating ChronoHive: The Pilot Process

How a ChronoHive pilot works: isolated and software-first, with methodology, arms, and success criteria fixed in writing before any run — and what you receive when it completes.

Version 1.0 · October 26, 2026 · PDF · 7 pages · Layer1Labs Silicon Inc.

Key figures

501

completed preregistered trials on real GPUs behind the published results, with every headline figure recomputed by an internal red team before publication

Published basis

+9.8%

end-to-end training speedup on real data — the figure a pilot sets out to reproduce or refute on your own workload

Measured — preregistered A/B trial

5–15%

the throughput range the model projects for pilot planning — a sensitivity range, not a production speedup

Modeled — not a measurement

At most 2 GPUs

the published measurement envelope: every result is single-node — no multi-node or cluster measurement exists yet

Published envelope

What it covers

  1. What an evaluation is
  2. First contact
  3. Scoping against the published envelope
  4. The preregistered pilot
  5. What you receive, and what follows
  6. Questions to settle in writing before the pilot
  7. The published limits

Basis

This guide describes the evaluation exactly as published on the ChronoHive page and the FAQ: licensing by written agreement, an evaluation path for qualified teams, and pilots that are isolated and software-first — run on the clusters you already own, with no hardware purchase.

What the site does not publish — timelines, costs, license terms — the guide does not invent. Those appear as questions to settle in writing during scoping, before the pilot’s methodology, arms, and success criteria are fixed.