Agent sandboxes / Collimate alternative

A Collimate sandbox alternative for durable work beyond rollout branches

Collimate focuses on live microVM forks for agents and RL rollouts. It already preserves execution state through suspend/resume and charges retained snapshot storage separately. Compare that branch-oriented workflow with a per-user agent computer whose filesystem grows across tasks.

Reviewed 2026-10-18 · Published by the Sandbox0 team

When Collimate fits

Collimate fits search trees, RL fan-out, and branching from a prepared running machine with live process memory. Its group-level fork API and independent parallel sandboxes address different forms of concurrency.

When to consider Sandbox0

Consider Sandbox0 when the central requirement is an evolving durable whole RootFS, automatic online growth, and supported persistent-file operations while compute is paused. Evaluate fork throughput and runtime compatibility independently of filesystem durability.

Capabilities

Collimate and Sandbox0: what persists?

Feature comparison with documentation and verification status
Environment and runtime lifetimeCollimateDemo is one hour, Pro 24 hours; Enterprise has no API session cap per current documentation.View evidence ↗Sandbox0ttl and hard_ttl default to 0 (disabled). Configure idle pause and deletion policies for the same persistent sandbox identity; an idle running sandbox still accrues compute charges.Read documentation →
Persistent filesCollimateSuspend retains saved sandbox state for resume; deletion reclaims it.View evidence ↗Sandbox0The writable RootFS survives successful checkpoints, including software, system configuration, dependencies, and user data.Read documentation →
Whole RootFSCollimateSuspend restores the saved filesystem and live process state; network identity changes on resume.View evidence ↗Sandbox0Persistence extends beyond /workspace to paths such as /usr/local, /etc, and /home. Runtime tmpfs and virtual filesystems are excluded.Read documentation →
Release compute, keep stateCollimateSuspend/resume preserves exact live state with separate retained snapshot storage.View evidence ↗Sandbox0Pause compute between tasks; resume on available capacity. Default resume starts fresh processes from the saved filesystem.Read documentation →
Memory and process stateCollimateLive forks retain process memory; suspend restores retained execution state.View evidence ↗Sandbox0Experimental memory checkpoints can preserve compatible process state. Use memory: true on both pause and resume on a compatible deployment. A filesystem checkpoint alone does not preserve running processes.Read documentation →

Online RootFS growth

On growth-enabled images, Sandbox0 automatically expands the writable RootFS as free space runs low, within a configured sparse logical ceiling up to 64 TiB. Agent processes keep running. Growth is enabled in the qualified Cloud region; confirm activation for other regions. Existing images and snapshots retain their captured storage geometry. The ceiling is logical capacity, not preallocated host disk.

Live-memory forks and suspend storage are useful persistence mechanisms, but the reviewed sources do not establish automatic online expansion of the complete mounted RootFS as one agent accumulates system tools and outputs.

Files while compute is paused

Sandbox0's Files API reads persistent files without resuming compute. Filesystem-only pauses also permit edits; retained-memory pauses are read-only and reject mutations with 409. SSH, SFTP, file watching, and runtime-only mounts require running compute.

See paused file operations →

Live forks and independent parallel machines have different placement

Collimate documents copy-on-write children that retain the parent’s live processes, while independent parallel environments can spread across the fleet. A live-fork group initially shares a node; it is not the same scheduling or fault boundary as separately provisioned machines.

Suspend retains state, but session and storage policies still apply

Suspend saves live state, releases RAM, and later resumes by ID. The limits guide caps Demo sessions at one hour and Pro at 24 hours, with uncapped Enterprise terms. Suspended snapshot storage remains a separate bill even though live compute stops. The suspend guide says resume restores the saved filesystem and process memory but assigns a fresh network identity.

Resolve the current concurrency entitlement before planning capacity

The public pricing page advertises Pro up to 100 concurrent sandboxes, while lifecycle and limits documentation describe uncapped Pro concurrency. Preserve that discrepancy rather than promise either figure universally. The current rate card separately prices held CPU/RAM and suspended snapshot bytes.

Pricing

Compare billing models, then price your workload

CPU allocation, memory, billing minimums, disk, plan fees, and supported workload differ. These meters are not equal-performance machine quotes. Confirm region, resource units, and included allowances before estimating a total.

Collimate

Pricing reviewed 2026-10-18

Live CPU
$0.040 per vCPU-hour
Held resources, metered per second.
Rate source ↗
Live RAM
$0.012 per GB-hour
Rate source ↗
Suspended snapshot storage
$0.10 per GB-month
Rate source ↗
Spawn and live fork
No separate operation charge
Children still accrue their own live resources.
Rate source ↗
Pro platform fee
No platform fee
Rate source ↗

Pro concurrency descriptions conflict between the price page and limits guide; confirm effective tenant limits. Preserve GB labels and distinguish suspended storage from running resources. Published fork-latency examples are not an independently measured workload guarantee.

Check current Collimate pricing ↗

Sandbox0

$0.015/GiB-hour compute (billed by the second at approximately $0.000004167/GiB-second), plus $0.02/GiB-month of retained incremental RootFS and memory checkpoint data. Shared base images and platform recovery retention are excluded; paused sandboxes accrue storage charges only.

Compute / GiB-hour
$0.015
Retained data / GiB-month
$0.02
Paused compute
$0.00

A 4 GiB agent, 100 running hours

$6.00 compute

4 GiB × 100 hours, billed by the second. Keeping 20 GiB of billable incremental data for a full billing month adds $0.40. CPU is derived from memory by the platform; choose a tier that fits your workload.

Understand storage billing →

Before migrating from Collimate

  1. Inventory running parents, fork groups, retained execution state, templates, saved files, session caps, and effective tenant concurrency.
  2. Export needed data and port compatible templates, commands, files, credentials, network rules, and rollout orchestration.
  3. Validate fork/workload behavior separately from repeated system persistence, RootFS growth, paused file access, and live plus suspended-storage cost.

Common questions

Is Sandbox0 a drop-in Collimate replacement?

Evaluate API compatibility and product scope. A shared sandbox category does not make SDKs, lifecycle semantics, or limits interchangeable.

Does keeping files also keep a running agent?

Filesystem state and volatile process state are separate. Sandbox0 requires explicit compatible memory checkpoints to preserve running processes; its default filesystem resume starts fresh processes.

Sources and review scope

The Boat team's Battleships directory is the discovery source. References below describe the product and plan discussed on this page. Check current availability, regional terms and billing units before migrating. Where the cited material does not establish a capability or price, the comparison identifies that uncertainty.

Related comparisons