Development environments / Moonshot Kimi Hosted Agents sandbox alternative

A Kimi Hosted Agents alternative

Kimi Hosted Agents combines the agent loop, versioned environments, plugins, credentials and a session sandbox. It already synchronizes working files and offers reusable memory stores. Compare those managed functions and selected-path retention with a runtime that preserves the whole writable Linux RootFS.

Start buildingCompare capabilities ↓

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

When Moonshot Kimi Hosted Agents sandbox fits

Kimi Hosted Agents fits applications that want managed model/tool execution, SSE session events, prepared environments and platform plugins. Its built-in document tooling and optional declared packages reduce the initial environment work for report and artifact tasks.

When to consider Sandbox0

Consider Sandbox0 when your own agent harness should return to installed system software, configuration and data across the writable RootFS, with online growth and supported paused files. Model routing, event history, plugins, memory services and agent orchestration remain application responsibilities.

Capabilities

Moonshot Kimi Hosted Agents sandbox and Sandbox0: what persists?

Feature comparison with documentation and verification status
Environment and runtime lifetimeMoonshot Kimi Hosted Agents 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 filesMoonshot Kimi Hosted Agents sandboxThe session’s synchronized /mnt/agents workspace is remounted after reconstruction; unsynchronized writes can be lost.View evidence ↗Sandbox0The writable RootFS survives successful checkpoints, including software, system configuration, dependencies, and user data.Read documentation →
Whole RootFSMoonshot Kimi Hosted Agents sandboxWorkspace sync and memory mounts do not establish durability of arbitrary changes across the complete writable Linux system.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 stateMoonshot Kimi Hosted Agents sandboxManaged session continuation does not by itself establish arbitrary VM execution-state pause/resume.View evidence ↗Sandbox0Pause compute between tasks; resume on available capacity. Default resume starts fresh processes from the saved filesystem.Read documentation →
Memory and process stateMoonshot Kimi Hosted Agents sandboxThe reviewed workspace and memory-store contracts concern files, not arbitrary live process memory.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.

Kimi documents synchronized session workspaces and built environment versions. These sources do not establish automatic online expansion or preservation of arbitrary system changes across the complete mounted RootFS. Sandbox0’s growth-enabled RootFS targets that persistent-computer requirement.

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.

Kimi exposes session-file browsing and downloading separately from uploaded-file resources. The reviewed file guide does not specify a no-wake full-RootFS mutation contract, so do not equate artifact retrieval with Sandbox0’s supported paused Files API.

See paused file operations →

Session workspace sync is useful but has a commit boundary

Files under /mnt/agents are synchronized into that session’s persistent workspace and remounted after sandbox reconstruction. Writes not yet synchronized can be lost on instance exit. Different sessions do not share that working directory; memory-store mounts provide an explicit cross-session path. Export critical results or publish an immutable artifact.

Source 1 ↗

An environment version is not the accumulated live machine

Kimi builds declared apt, pip, npm, cargo and Go dependencies plus a setup script into a reusable environment. A session pins the latest ready version at creation; updates affect future sessions. This reproduces intended setup without proving that arbitrary system changes made during a task persist with the workspace.

Source 1 ↗

Network policy and platform tools have separate scope

Environment networking governs sandbox processes, with unrestricted as the default and a limited host policy available. The guide explicitly excludes platform tools such as Web Search/Fetch from that process policy. Preserve that distinction when moving tool execution and network controls into your own runtime architecture.

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.

Moonshot Kimi Hosted Agents sandbox

Pricing reviewed 2026-10-18

Hosted cloud sandbox
CNY ¥0.072/hour
Published shape: one CPU core and 1 GB RAM; runtime billing with second-level minimum.
Rate source ↗
Model inference
Selected model’s separate token rates
Sandbox runtime does not include all model inference.
Rate source ↗
Web-search plugin
CNY ¥0.03 per use
Other plugins have their own per-call rates.
Rate source ↗

Keep the official CNY denomination and GB label; no assumed USD exchange rate or equal-throughput claim is used. A complete task can incur sandbox time, model tokens and multiple plugin charges. The runtime price is not an all-inclusive agent-session flat fee.

Check current Moonshot Kimi Hosted Agents 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 Moonshot Kimi Hosted Agents sandbox

  1. Export important session files and immutable artifacts, memory-store content, environment definitions, agent instructions and plugin/credential references.
  2. Run a compatible agent harness on Sandbox0, recreating model calls, SSE/application events, tools, credential handling and network policy explicitly.
  3. Test system-package changes outside /mnt/agents, RootFS growth, committed-state recovery, paused files and combined model/plugin/runtime cost.

Common questions

Is Sandbox0 a drop-in Moonshot Kimi Hosted Agents sandbox replacement?

Evaluate API compatibility and product scope. These products serve different core workflows, so a change may require a different architecture.

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 ↗platform.kimi.com/docs/hosted-agents/sandbox-reference.md ↗platform.kimi.com/docs/hosted-agents/environments.md ↗platform.kimi.com/docs/hosted-agents/quickstart.md ↗platform.kimi.com/docs/pricing/hosted-agents.md ↗Sandbox0 RootFS capacity →

Related comparisons

Agency Tool Company vs Sandbox0 →Amp Orbs vs Sandbox0 →Bitrise RDE vs Sandbox0 →boxes.dev vs Sandbox0 →Capy vs Sandbox0 →Claw 2 Agent vs Sandbox0 →