Agent sandboxes / Tangle Sandbox alternative

A Tangle sandbox alternative for durable machine state

Tangle separates a sandbox from the agent conversations inside it, and supports several execution drivers. That distinction matters for returning agents: durable conversation history is useful, but it is different from keeping the entire changed Linux machine.

Start buildingCompare capabilities ↓

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

When Tangle Sandbox fits

Tangle fits teams that want managed agent sessions, conversation branching, fleets, and a common SDK over container, microVM, or operator-backed execution. Choose its driver and agent mode based on the actual isolation and workflow requirements.

When to consider Sandbox0

Consider Sandbox0 when the central requirement is a durable writable RootFS for an agent you control, with online growth, lifecycle controls for long work, and supported file access during compute pauses. Port the surrounding agent orchestration separately.

Capabilities

Tangle Sandbox and Sandbox0: what persists?

Feature comparison with documentation and verification status
Environment and runtime lifetimeTangle SandboxNot established in this reviewSandbox0ttl 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 filesTangle SandboxFilesystem snapshots are documented; verify the selected driver, mount coverage, and retention terms.View evidence ↗Sandbox0The writable RootFS survives successful checkpoints, including software, system configuration, dependencies, and user data.Read documentation →
Whole RootFSTangle SandboxFilesystem snapshots do not establish uniform complete RootFS coverage across every driver and mount.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 stateTangle SandboxThe SDK documents sandbox stop/resume; that alone does not establish retained execution memory.View evidence ↗Sandbox0Pause compute between tasks; resume on available capacity. Default resume starts fresh processes from the saved filesystem.Read documentation →
Memory and process stateTangle SandboxConversation forks and filesystem snapshots are not evidence of RAM/process checkpointing.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.

A conversation fork or a filesystem snapshot is not automatic expansion of the live complete RootFS. The reviewed SDK does not establish that growth contract uniformly across Tangle’s execution drivers.

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 →

Conversation branching is different from machine forking

Tangle’s session fork copies conversation history and can isolate agent work with Git worktrees. Its filesystem snapshot and sandbox lifecycle APIs describe different objects. A forked conversation does not establish a full machine clone or a retained RAM checkpoint; map each application artifact to the correct persistence boundary.

Source 1 ↗

Driver and agent mode determine the effective surface

Tangle documents container, microVM, and operator-backed execution, rather than one universal isolation implementation. Bare execution offers process/lifecycle access, while agent-disabled managed mode retains additional tooling without a default agent. The lifetime and idle fields are configurable, but example values are not proof of a product-wide maximum.

Source 1 ↗Source 2 ↗

Reference rates need a selected pool and effective plan

The public general-pool pricing endpoint marks its values as reference constants. Its hourly floor is a rate used in per-second metering, not a one-hour minimum session. Egress has its own plan, scope, and allowance: the free allowance is account-wide and refuses excess, while paid allowances are per sandbox.

Source 1 ↗Source 2 ↗Source 3 ↗

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.

Tangle Sandbox

Pricing reviewed 2026-10-18

General pool CPU reference
About $0.010446 per core-hour
Rounded public reference constant; confirm the effective worker quote.
Rate source ↗
General pool RAM reference
About $0.007051 per GB-hour
Rate source ↗
General pool disk reference
About $0.000734 per GB-hour
Rate source ↗
Minimum running rate
$0.04 per hour, metered per second
An hourly rate floor, not a minimum one-hour duration.
Rate source ↗
Pro outbound overage
$0.05 per decimal GB beyond 2 TB per sandbox per calendar month
The Free plan instead includes 100 GiB per account and refuses overage. Plan fees are not verified here.
Rate source ↗

These public API rates are reference values, not a verified bound-worker quote or complete plan invoice. Resource charges use the higher of the resource sum and running-rate floor. Egress is outbound, uses 1,000,000,000 bytes per billed GB, and has separate account versus sandbox scope. Prepaid owner balance and any plan or agent/model charges need separate confirmation.

Check current Tangle Sandbox 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 Tangle Sandbox

  1. Inventory sandbox drivers, agent conversations, Git worktrees, filesystem snapshots, mounts, credentials, and current pool/plan terms.
  2. Export required filesystem state and conversation history independently, then port the agent runner and machine command/file interfaces.
  3. Validate system-package persistence, growing RootFS data, paused file access, process-continuity requirements, and charges including the running-rate floor and scoped egress.

Common questions

Is Sandbox0 a drop-in Tangle Sandbox 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 ↗sandbox.tangle.tools/docs/sdk/typescript ↗sandbox.tangle.tools/llms.txt ↗sandbox.tangle.tools/v1/pricing/sandboxes ↗sandbox.tangle.tools/v1/pricing/egress ↗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 →