Agent sandboxes / Beam alternative

A Beam Cloud sandbox alternative for a reusable agent computer

Beam supports both reusable root-filesystem images and memory snapshots on supported sandbox runtimes. The important questions are how the agent returns to saved state, which mounts have separate persistence, and which Beam product meter applies.

Start buildingCompare capabilities ↓

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

When Beam fits

Beam fits teams combining sandbox execution with its Python infrastructure, GPU tasks, volumes, and inference or queue workloads. Choose its sandbox, VM, and serverless products deliberately because their prices and lifecycle primitives differ.

When to consider Sandbox0

Consider Sandbox0 when one per-user environment should keep its accumulated RootFS across tasks, grow without a disk-size replacement, and expose persistent files during compute pauses.

Capabilities

Beam and Sandbox0: what persists?

Feature comparison with documentation and verification status
Environment and runtime lifetimeBeamTTL defaults to 600 seconds; -1 requests explicit application termination.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 filesBeamFilesystem images retain captured files; volumes and durable disks are separate state sources.View evidence ↗Sandbox0The writable RootFS survives successful checkpoints, including software, system configuration, dependencies, and user data.Read documentation →
Whole RootFSBeamReusable root-filesystem images; mounted datasets need separate retention.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 stateBeamThe reviewed recovery path restores a snapshot into a new sandbox identity and URLs.View evidence ↗Sandbox0Pause compute between tasks; resume on available capacity. Default resume starts fresh processes from the saved filesystem.Read documentation →
Memory and process stateBeamSupported runtimes can snapshot process state and filesystem.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.

Beam offers several persistence primitives, including snapshot-derived root filesystems and mounted durable disks. The reviewed sandbox configuration does not establish automatic growth of its running root filesystem; compare the chosen product and mount layout explicitly.

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 →

A filesystem image and a memory checkpoint differ

Beam’s filesystem-image API produces an image for new sandboxes. Its memory-snapshot API captures process state and filesystem on supported runtimes and restores into a new identity with new URLs. Keep those artifact IDs and independent mounted data separate during recovery.

Source 1 ↗

A running command does not extend a finite TTL

Beam documents keep_warm_seconds as a sandbox TTL, not an idle timeout reset by each command. The default is 600 seconds; -1 leaves termination to the application. Use explicit TTL updates for finite tasks so an agent does not lose its runtime while still working.

Source 1 ↗

Use the sandbox-specific CPU price

Beam publishes different meters for VMs, CPU-only serverless tasks, GPU-attached tasks, and sandboxes. Sandbox CPU uses physical cores, with two vCPUs per core. Applying the cheaper VM or serverless row to an interactive sandbox would understate its rate.

Source 1 ↗

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.

Same size. Lower compute cost.

2 vCPU · 4 GiB RAM · 730 running hours / month

Save $122.03 / month

The same CPU and memory allocation, with runtime aggregated across sessions. Monthly allowances and credits below are applied to this workload.

Beam

Sandbox usage billing

$165.83

$0.22716/hour before monthly allowances

Sandbox0

Pay as you go · no monthly plan fee

$43.80

$0.06/hour · 73% lower in this example

Uses Beam’s Sandbox rates, with one physical core equivalent to two vCPUs. Its separate VM and Serverless rates are different product meters; persistent storage is outside this compute estimate.

Compute and stated plan fees only. Storage, network, taxes and one-time credits are excluded; hardware and throughput can differ.

Calculation and billing assumptions

Beam: $165.8268 compute after 0 included vCPU-hours and 0 included GiB-hours, plus $0.00 plan fee, minus $0.00 applied monthly credit.

Sandbox0: 4 GiB × 730 hours × its published compute rate. CPU uses the Cloud allocation shown on the pricing page.

Includes compute and the selected plan. Excludes storage, network, taxes, one-time grants and negotiated discounts. Assumes this workload can use the full listed monthly allowances. CPU and memory counts do not guarantee equal hardware or throughput.

Checked Beam rate source ↗Sandbox0 rates ↗

Beam

Pricing reviewed 2026-10-18

Sandbox CPU
$0.0000375 per physical-core-second
One physical core is equivalent to two vCPUs.
Rate source ↗
Sandbox memory
$0.0000064 per GiB-second
Rate source ↗
Persistent volumes above included storage
$0.021 per GB-month
The published storage table includes the first 1 TB and snapshots.
Rate source ↗

Sandboxes are billed by the millisecond while running. These are sandbox rates; Beam’s persistent VM and serverless prices are separate product rows. Storage, resource shape, and supported snapshot runtime must match the actual workload.

Check current Beam 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
Understand storage billing →

Before migrating from Beam

  1. Export filesystem-image IDs, memory-checkpoint IDs, shared volumes, durable disks, and buckets as distinct state sources.
  2. Replace TTL management and new-URL handling with the intended Sandbox0 sandbox lifecycle and ingress configuration.
  3. Measure a representative task using the actual CPU entitlement and verify RootFS growth, pause, offline file access, and restart semantics.

Common questions

Is Sandbox0 a drop-in Beam 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.

Battleships research record ↗docs.beam.cloud/v2/sandbox/snapshots ↗docs.beam.cloud/v2/sandbox/configuration ↗www.beam.cloud/pricing ↗Sandbox0 RootFS capacity →

Related comparisons

Agent 37 vs Sandbox0 →AgentComputer vs Sandbox0 →ainclave vs Sandbox0 →Alibaba Cloud Agent Sandbox / FC / AgentRun vs Sandbox0 →Alibaba Cloud AgentBay vs Sandbox0 →Amika vs Sandbox0 →