Linux sandboxes for agents

Fast sandboxes.Normal compute prices.

Run production coding workloads on hardware-isolated Linux machines for a fraction of what the usual sandbox providers charge.

2.6¢per active hour2 vCPU · 4 GiB

Calculate your savings See the benchmark

Normalized hourly compute2 vCPU / 4 GiB

Run Cloud1×$0.026

Fly.io3.7×$0.098

CodeSandbox5.7×$0.150

E2B / Daytona / Blaxel6.3×$0.166

Cloudflare6.8×$0.180

Modal9×$0.237

Vercel12.9×$0.341

Published usage rates or nearest matching tier, reviewed July 12, 2026. Plan fees, credits, storage, network, and creation charges are excluded.

2.6¢ / hour

2 vCPU and 4 GiB at launch rates

Would rank #1*

Unofficial Dax benchmark result

Per-second billing

Pause with no compute charge

Real workload performance

Cheap because the infrastructure is efficient.Not because the machines are slow.

ComputeSDK's Dax benchmark exercises a complete coding workload inside a fresh sandbox—not merely an API response or warm-start timer.

Read the public methodology

Dax / OpenCode workloadFull run

01Prepare Linux

02Install Node + Bun

03Clone OpenCode

04Install packages

05Typecheck

Run Cloud result

unofficial*

Highest score in our run of the public benchmark.

*Run Cloud's result was produced by us using the benchmark's public script and is not yet included in ComputeSDK's official provider results.

Your workload, your bill

Spend your budget on the agent.Not the sandbox markup.

Set your concurrency and monthly runtime to compare the same normalized Linux workload across providers.

Concurrent sandboxes50Runtime per sandbox / month240 hr / 10 days

Region selection

Broad and exact placement apply only published region-selection premiums. Providers without a published multiplier remain unchanged.

Normalized to 2 vCPU / 4 GiB. Estimates apply each provider's published usage rates or nearest matching tier to the same runtime.

Run Cloud1×

Per-second CPU + memory

$316.51

Fly.io≈3.7×

Performance-2x / 4 GB regional rate

$1,172.40

CodeSandbox≈5.7×

On-demand VM credit rate

$1,800.00

E2B≈6.3×

Per-second CPU + memory

$1,987.20

Daytona≈6.3×

Per-second CPU + memory

$1,987.20

Blaxel≈6.3×

Active compute by allocated memory

$1,987.20

Cloudflare≈6.8×

Containers at 100% active CPU

$2,160.00

Modal≈9×

Sandbox rate at 3× base CPU + memory

$2,848.61

Vercel≈12.9×

Sandbox at 100% active CPU

$4,089.60

Illustrative usage estimate only. Plan fees, included credits, storage, network, and creation charges are excluded. Active-CPU providers assume 100% utilization; lower utilization can reduce their actual bill.

Lower switching cost

Change your provider.Not your product.

Keep the lifecycle your application already understands and move workloads over incrementally with provider-shaped compatibility adapters.

Migrating from another sandbox platform?

Keep your existing lifecycle code and move workloads over incrementally with a provider-specific compatibility adapter.

Example adapter · Modal

compat_modal.py

# from modal import Sandboxfrom run_cloud.compat.modal import Sandbox # The rest of your code stays the same.

Compatibility matrix

One runtime. Eight migration paths.

Provider-shaped entry points let you redirect existing integrations to Run Cloud with an import change instead of an orchestration rewrite.

Swipe to compare →

Sandbox provider compatibility adapters available for Run Cloud
Provider	Language	Original package	Run Cloud adapter	API shape	Migration	Guide
01Modal	Python	modal	run_cloud.compat.modal	Sandbox	Import change	Compare →
02E2B	Python	e2b	run_cloud.compat.e2b	Sandbox	Import change	Compare →
03Blaxel	Python	blaxel	run_cloud.compat.blaxel	Sandbox	Import change	Compare →
04CodeSandbox	TypeScript	@codesandbox/sdk	@run-cloud/sdk/compat/codesandbox	Sandbox	Import change	Compare →
05Vercel	TypeScript	@vercel/sandbox	@run-cloud/sdk/compat/vercel	Sandbox	Import change	Compare →
06Cloudflare	TypeScript	@cloudflare/sandbox	@run-cloud/sdk/compat/cloudflare	Sandbox	Import change	Compare →
07Fly.io	TypeScript	@flydotio/machines	@run-cloud/sdk/compat/fly	Machines	Import change	Compare →
08Daytona	Python	daytona	run_cloud.compat.daytona	Daytona	Import change	Compare →

Core provider entry points are preserved by each adapter.8 providers · Python + TypeScript

No missing pieces

Everything you expected from an expensive sandbox.

A hard isolation boundary

Every workload runs inside a KVM-backed Firecracker microVM with its own kernel and filesystem.

Real Linux, with root

Use apt, curl, Git, language runtimes, package managers, and the tools your coding workloads already expect.

State that can branch

Snapshot an environment, restore it later, or fork it into parallel agent work without starting over.

Pause at no compute cost

Preserve useful state between agent turns without paying for an idle machine.

Ready in under a second

Keep interactive agent work moving with a warm create path designed for sub-second startup.

One machine or hundreds

Fan out evaluations, builds, and agent branches without maintaining a standing fleet.

More than Linux when you need it

Let the same agent continue onto an iPhone.

Add browser-controlled iOS simulators and cloud Xcode builds when the work needs to leave the Linux sandbox.

Explore iOS simulators

FAQ

The practical details.

How much does a Run Cloud sandbox cost?

A 2 vCPU and 4 GiB Linux sandbox costs approximately $0.026376 per active hour at current launch rates. CPU and memory are metered independently by the second.

What does the Dax benchmark measure?

It runs a cold clone, install, and typecheck cycle for the OpenCode repository inside a fresh sandbox. The workload prepares Linux, installs Node and Bun, clones a pinned commit, installs packages, and runs the project typecheck.

Is the Run Cloud benchmark result official?

Not yet. The Run Cloud result was produced using the benchmark’s public script but is not currently part of ComputeSDK’s published provider runs. It is presented as an unofficial result until an independent run is published.

Can I migrate an existing sandbox integration?

Run Cloud provides provider-shaped compatibility adapters for common sandbox APIs. They are intended to preserve core lifecycle code while you migrate incrementally and validate provider-specific behavior.

Are these containers or virtual machines?

Run Cloud Linux sandboxes are KVM-backed Firecracker microVMs with a dedicated kernel and filesystem rather than shared-kernel containers.

See what your sandbox workload should cost.

Compare the bill, test the workload, and migrate without rebuilding your product around another provider.

Get early access Calculate your savings