The short version

Editorial overview — a catalog-level terms check. Everything below was verified against Moli and Lexmount’s own published pages on 2026-09-21 — the Lexmount pricing page, which declares itself the “public product catalog for review”, and the public GitHub repository (license badge and README) — with no account created and nothing run in the browser, so there is no hands-on claim and no star rating on this page. The plan tables are quoted from those two sources; this page exists because the catalog and the open-source repo each tell only half the story.

Two products share the name, and the pricing story is mostly about the difference between them:

  • Moli (open source) — github.com/lexmount/moli, Apache-2.0, written in Rust: a browser engine for automation that skips the visual render and treats the DOM + style state as ground truth (README, captured 2026-09-21). Same-day API: 2,164 stars / 146 forks / pushed 2026-09-21. Free as in license, as in price, as in redistribution — genuinely permissive, the opposite end of the spectrum from n8n’s fair-code (see Is n8n open source?).
  • Lexmount cloud (hosted) — browser.lexmount.com/pricing metering browser-hours and fetch requests per month, $0 to $180.

That second bullet is where the page gets interesting.

The pricing page that says “buy elsewhere”

Verbatim notice at the top of the captured cloud page:

“This page is the public product catalog for review. Online payment is not enabled and Lexmount does not collect payment here yet.”

And below the fold: “PayPal checkout is pending merchant approval”. Every paid card’s button is “Contact sales” — only the $0 Free tier has a real “Start free”. So: published prices, no way to pay them on the page, today. The plan structure (below) is quoted as catalog truth; treat availability as pre-checkout.

Cloud plans (USD, captured 2026-09-21, sold as one-time · 1-month units)

Plan Price Browser hours/mo Web Fetch/mo Concurrency Persistent contexts Proxy traffic
Free $0 1 h 1,000 3 0 0 B
Pro $12 100 h 1,000 25 10 100 MB
Pro+ $60 500 h 10,000 100 50 1 GB
Enterprise $180 1,500 h 10,000 250 200 1 GB

Billing mechanics printed in the FAQ: “One-time purchases do not auto-renew” — this is not-a-subscription in the truest sense we’ve catalogued — and a full refund is available “within 7 calendar days only while all attributable paid credits remain unused”. The page’s own worked example: Pro $12.00 + a 100,000-request fetch pack $47.00 = $59.00, pack explicitly “one-time, not a recurring monthly charge”.

Web Fetch packs (per 10k requests, by your plan rate): Pro $4.70, Pro+ $4.30, Enterprise $3.90 — i.e. $47 / $43 / $39 per 100,000, $470/$430/$390 per million. Purchased balance does not reset monthly and survives a downgrade to Free, but Free accounts cannot buy new packs.

The honest caveat: at $12 for 100 hours, cloud Moli undercuts hosted browser-automation competitors on sticker price — but “you can’t actually check out yet” is a bigger adoption risk than any number in the table.

The kernel claim, and its opt-in trap

Two README statements worth quoting for anyone planning real work on this:

“What most browser automation tasks actually need is page structure, not a continuously rendered visual world. Moli treats the native DOM and style state as the single source of truth, triggering layout or software paint only for operations that genuinely require them.”

That’s the pitch: an independent Rust engine (V8 + its own layout/paint path), not a headless Chromium wrapper — and the README shows Playwright-compatible usage via chromium.connectOverCDP, which is how agent stacks would actually consume it.

And the trap, verbatim:

“Expensive browser operations in Moli require an explicit opt-in and are never enabled by default.”

Rendering, screenshots, and other paint-dependent operations fail closed until you turn them on. Good default for cost control, a footgun for anyone porting Chromium scripts expecting pixels on demand.

Ownership question, closed

Earlier research (2026-09-16/21) flagged a possible org migration: a moli-browser/moli- repo mirroring the README, and star counts disagreeing between 1,760 and 2,016. Same-day check settles it: api.github.com/repos/moli-browser/moli → 404; lexmount/moli → live, 2,164 stars, pushed 2026-09-21. lexmount/moli is the canonical repo; cite it.

What we did not verify

  • We ran nothing: no local Moli install, no cloud instance (and the cloud cannot be purchased on-page today). Speed/compatibility claims are README self-descriptions, flagged as such.
  • The “1 browser hour” Free tier: we did not test what consumes hours or how they meter.
  • No refund/support SLA testing; the 7-day/unused-credits rule is quoted, not exercised.
  • The hosted product’s terms, data policy, and seller entity disclosures sit behind “Review the seller… terms” links we did not crawl in this pass.

Verdict (terms-level, no score)

No rating — nothing was installed or run. What the terms support: a permissively-licensed, actively-pushed Rust engine plus an unusually transparent, non-auto-renewing price list with worked examples and honest billing mechanics. What they cannot support: buying it. A catalog with payment disabled is pre-commercial, and the $12 entry price should be read as a posted intent, not a transactable fact, as of 2026-09-21.

Methodology & sources

  • browser.lexmount.com/pricing — captured 2026-09-21 (UTC) in a proxied Chromium session; anchors $12/$60/$180, $47/$43/$39, “concurrent”, “refund”, “not enabled” all matched in the rendered text; DOM dump archived at eval-ops/runs/2026-09-21-group8/moli-cloud-pricing.txt.
  • github.com/lexmount/moli page + GitHub REST API — same-day: 2,164 stars, 146 forks, 133 open issues+PRs, Apache-2.0, pushed_at 2026-09-21T09:28:14Z; moli-browser/moli API probe → 404.
  • README quotes (architecture, opt-in, CDP example) from the same 2026-09-21 page capture (moli-repo.txt).
  • Related: Is n8n open source? and n8n alternatives for the license-spectrum context this tool’s Apache-2.0 sits at the free end of.
  • Checked 2026-09-21. Pricing pages change; re-verify before citing.