Agent Sandboxesagentsandboxes.org
Sandbox profile · Open-source project

Celesto (SmolVM)

Answer in brief

Open-source microVM "computers" for agents from Celesto AI, published as the smolvm package, with Firecracker on Linux and QEMU on macOS, plus a hosted cloud option. A young, alpha-labelled but feature-rich open-source microVM runtime that is easy to try locally on macOS or Linux and has a per-hour cloud behind it.

By the Agent Sandboxes editors · Updated

Overview

Celesto is an Apache-2.0 runtime from Celesto AI that gives an agent a "secure and persistent computer": a microVM booted with Firecracker on Linux or QEMU on macOS (Apple Silicon), driven from a Python SDK (from celesto import Computer), a TypeScript SDK, a celesto CLI and an MCP server. The project was published as SmolVM (the smolvm package on PyPI, 0.0.36 in September 2026, classified "3 - Alpha") and the GitHub repository CelestoAI/SmolVM now resolves to CelestoAI/celesto. Computers can run a browser session with a CDP endpoint, a desktop with GUI apps, and Windows guests on a Linux host with KVM.

State is a documented focus: snapshots capture memory, disk and processes, and persistent state survives across sessions. The same Computer API runs locally or against Celesto's cloud with provider="cloud", billed per hour from $0.015 for a 1 vCPU/2 GB Micro computer. Maturity labels in the sources: "3 - Alpha" on PyPI, "alpha" for the TypeScript SDK, "preview" for the macOS desktop and Windows guests. The company also sells GitHub Actions runners on the same infrastructure.

Notable

  • PyPI package smolvm 0.0.36 released 17 September 2026 with development status "3 - Alpha"; TypeScript SDK labelled alpha
  • README claims a computer "starts in about 500 ms"; the website claims hosted sandboxes "start in under 2 seconds"
  • Hosted computers from $0.015 per hour (Micro, 1 vCPU/2 GB) with a one-time $5 credit

Celesto (SmolVM) specifications

AttributeDetails
IsolationMicroVM per computer with hardware-level separation from the host: Firecracker on Linux (KVM), QEMU on macOS; the PyPI listing also names libkrun as a backend
DeploymentOpen-source project · Self-hostable
Open sourceYes (Apache-2.0)
SDKs and interfacesPython, JavaScript/TypeScript, CLI, MCP server
Cold start (vendor claim)README: a local computer "starts in about 500 ms"; the website says hosted agent sandboxes "start in under 2 seconds". Docs note QEMU on macOS is 3-4x slower than Firecracker for lifecycle operations
Max session lengthNot documented
PersistenceSnapshots "pause and restore memory, disk, and active processes"; persistent state survives across sessions; in the cloud, stopped computers keep files, packages and workspace state, and stopped root disks are kept free for 30 days
GPUNot documented
NetworkingNetwork policy controls documented for Linux with Firecracker only; the QEMU backend uses user-mode (slirp) networking with port forwarding; cloud computers can expose public ports
PricingCloud computers billed per hour of running time, prepaid with no monthly fee: Micro 1 vCPU/2 GB $0.015/h, Small 1 vCPU/4 GB $0.020/h, Standard 2 vCPU/4 GB $0.030/h, Memory 2 vCPU/8 GB $0.040/h, Large 4 vCPU/16 GB $0.080/h, XL 8 vCPU/32 GB $0.160/h; root disk included; up to 10 running sandboxes. Local runtime is free
Free tierOne-time $5 Micro runtime credit, no card, expires 30 days after activation (one retained Micro sandbox); free to self-host
ComplianceNot documented
RegionsNot documented

Facts from Celesto (SmolVM)’s public documentation, verified November 9, 2026. Performance numbers are vendor claims.

Strengths and trade-offs

Strengths

  • You want a microVM runtime that works on an Apple Silicon laptop (QEMU) and on Linux (Firecracker) behind one API
  • You need snapshots that capture memory and running processes, not just a disk image
  • You want a cheap hosted fallback with the same SDK when a local machine is not available
  • You are building computer-use agents and want a browser or desktop inside the VM

Trade-offs

  • Alpha-labelled release line; the project name changed from SmolVM to Celesto and documentation is split across the old and new names
  • Network policy controls are Linux/Firecracker only; the macOS QEMU backend is documented as 3-4x slower for lifecycle operations
  • No compliance, region or session-limit documentation

Good fit for: Running agent computers as microVMs on your own Linux or Apple Silicon machine, with the same API available as a hosted cloud computer.

How Celesto (SmolVM) compares

SandboxIsolationOpen sourceSelf-host / BYOCCold start (vendor)Max sessionSnapshots / pauseGPUSDKs
Celesto (SmolVM)Firecracker / QEMU microVMYesYes / No~500 ms (claim)Not documentedYes / YesNot documentedPython, JavaScript/TypeScript, CLI, MCP server
BoxLitemicroVM + jailerYesYes / No~1 s (claim)Not documentedNot documented / Not documentedNot documentedPython, JavaScript/TypeScript, Go, Rust, C, CLI, REST API
MicrosandboxMicroVM (libkrun)YesYes / Yes<100 msNot documentedYes / Not documentedNoTypeScript, Python, Rust, Go, Ruby, CLI, MCP server
ArrakisCloud Hypervisor microVMYesYes / NoNot documentedNot documentedYes / Not documentedNot documentedPython, REST API, CLI, MCP server
FirecrackerKVM microVMYesYes / No<125 ms (spec)Not documentedYes / YesNoREST API, CLI
E2BFirecracker microVMYesYes / YesNot documented1 h / 24 h (Pro)Yes / YesNoPython, JavaScript/TypeScript
agents-sandboxmicrosandbox microVMYesNo / NoNot documentedNot documentedNot documented / Not documentedNot documentedCLI

See all Celesto (SmolVM) alternatives →

Frequently asked questions

What is Celesto (SmolVM)?
Open-source microVM "computers" for agents from Celesto AI, published as the smolvm package, with Firecracker on Linux and QEMU on macOS, plus a hosted cloud option. A young, alpha-labelled but feature-rich open-source microVM runtime that is easy to try locally on macOS or Linux and has a per-hour cloud behind it.
How does Celesto (SmolVM) isolate code?
Celesto (SmolVM) documents its isolation as: MicroVM per computer with hardware-level separation from the host: Firecracker on Linux (KVM), QEMU on macOS; the PyPI listing also names libkrun as a backend.
Is Celesto (SmolVM) open source?
Yes (Apache-2.0). It can be self-hosted.
How much does Celesto (SmolVM) cost?
Cloud computers billed per hour of running time, prepaid with no monthly fee: Micro 1 vCPU/2 GB $0.015/h, Small 1 vCPU/4 GB $0.020/h, Standard 2 vCPU/4 GB $0.030/h, Memory 2 vCPU/8 GB $0.040/h, Large 4 vCPU/16 GB $0.080/h, XL 8 vCPU/32 GB $0.160/h; root disk included; up to 10 running sandboxes. Local runtime is free Check the official pricing page for current rates.
What are the alternatives to Celesto (SmolVM)?
Common alternatives include BoxLite, Microsandbox, Arrakis, Firecracker, E2B, agents-sandbox. See Celesto (SmolVM) alternatives for a full comparison.

Sources

  1. celesto.ai
  2. celesto.ai/docs/
  3. celesto.ai/docs/cloud/overview.md
  4. celesto.ai/pricing
  5. github.com/CelestoAI/SmolVM
  6. pypi.org/project/smolvm/

Cite this page

Agent Sandboxes. (2026). Celesto (SmolVM). agentsandboxes.org. Last reviewed November 9, 2026. https://agentsandboxes.org/sandboxes/smolvm/

Found an error or a missing sandbox? Submit a correction — every entry links to the sources it relies on.