How we checked — Sandbox comparison

What was done

For each provider, a researcher read the documentation, the pricing page, the FAQ, the terms and, where it existed, the machine-readable OpenAPI or documentation bundle. Each profile records the exact quote, its URL and a tag. The profiles live beside the site inresearch/.

No account was created, no sandbox was started and nothing was billed. Every statement about behaviour is a documentation claim, not a measurement. Where a provider's pages contradict each other, both statements are shown and the row is marked as a contradiction.

Confidence tags

Tag	What it means
docs-confirmed	The provider's own documentation says it at the linked page.
pricing-page	The provider's pricing page or a money-specific page says it.
third-party	Someone other than the provider said it, such as battleships.dev or a review.
inferred	Our reading of documented facts. Not a sentence the provider wrote.
not-published	We looked and found no statement. We do not guess.

Dates

All twelve profiles were checked on 2026-10-28. Prices changed during October 2026 at several providers — boxd cut its rates by about 20% between 9 and 28 October, and Boat moved from credit packs to a dollar wallet — so treat any number as a dated observation, not a standing fact. Boat's documentation is served through a cache that can return an older revision, so its figures were re-fetched with cache-busting URLs.

Limits worth stating plainly

This is desk research. No provider account was created and no sandbox was started or billed, so every claim about behaviour is a documentation claim, not something we measured.
Prices are the providers' published list rates in US dollars on 2026-10-28, excluding tax. Several providers changed prices during October 2026, so a number here can be out of date within weeks.
Where a provider publishes nothing, the site says 'not published' rather than estimating. Where two pages of the same provider disagree, the site shows both and marks it as a contradiction.
battleships.dev is a comparison dataset built by the Boat (boat.dev) team. We used it as a cross-check, never as a primary source, and it disagreed with live provider pages often enough that its own numbers are not repeated here as fact.

About battleships.dev

battleships.dev is a comparison dataset built by the Boat team. We used it as a cross-check and never as a primary source. It disagreed with live provider pages often enough that its numbers are not repeated here as fact: it described InstaBox's pricing and sizes as they were before 20 October, boxd's rates as they were on 14 October, and Runtime's idle default as it was before 18 October. Its own Boat record is dated 30 September and disagrees with Boat's current docs on disk sizes and snapshot retention.

That does not make it useless. It is a fast way to find which questions to ask, and the right place to price your own workload. It is a bad way to settle a claim about behaviour.

Open questions we could not settle

Why Sandboxer's sizes, hourly rates and plan tiers match Boat's exactly while its disk sizes, regions, stopped-disk retention and legal entity differ. No public source explains the relationship.
Whether Agent 37's Cloud dashboard can create a sandbox by clicking. No documented path exists, but the absence may be a documentation gap.
Whether InstaBox's free plan requires a card. The pricing page says no; the API has a state that says yes.
What a stored snapshot costs at Lizard, and what backup storage costs at boxd. Neither is published.
Whether egress is free at Agent 37, boxd and Boat. None publishes a rate or a cap.
Morph's regions, maximum machine sizes and GPU availability. Its terms name the company (Morph Labs, Inc., San Francisco), but none of these is published, and two third-party sources disagree on whether it has raised funding.

Coverage

12 providers and 16 features. Every feature page carries a row for every provider, including “not published” where the provider is silent.