CodeSandbox SDK vs Replit
CodeSandbox SDK is a sandbox API: your code starts Firecracker VMs that hibernate with their memory and wake on the next request. Replit is an AI app builder: you describe an app, Replit Agent writes it, and Replit hosts it with Postgres and storage.
The companies and products
CodeSandbox SDK
CodeSandbox SDK is an API from CodeSandbox, now a Together AI company, that starts sandboxed development VMs on the same infrastructure as the CodeSandbox editor.
Each sandbox is a Firecracker microVM built from a template or any Dockerfile. When it goes idle, it hibernates with a memory snapshot, and it resumes or forks from that snapshot in seconds. You drive it from a TypeScript SDK through terminals, commands, files and HTTPS hosts. Together's docs call it the deprecated offering and point new users to its successor, the together-sandbox SDK.
Replit
Replit is an AI app builder and hosting platform: you describe an app in chat, Replit Agent writes it, and Replit publishes it.
You work in a browser workspace where Replit Agent sets up the project, writes the code, tests it and fixes problems, with system packages managed by Nix. When the app is ready you publish it as an Autoscale, Static, Scheduled or Reserved VM deployment, with Postgres and object storage next to it. It is built for people and teams who want a working app, not for programs that need to create many machines by API. This page compares the parts where the two overlap.
Which use cases each one fits
For each use case, Battleships prices every sandbox provider on the same workload and sorts them from the cheapest monthly bill up. “32nd cheapest of 365” means 31 of the other 364 providers run that workload, with every feature it needs, for less. “With limits” means the provider runs it but misses something, so it is listed after every full match. Open a use case for its whole ranking and what each provider misses.
| Use case and its workload | CodeSandbox SDK: price rank | Replit: price rank | boat: price rank |
|---|---|---|---|
| Software factory 100 at once · 4 vCPU / 8 GB · 2,200 sessions of 8 h a month A software factory starts and steers many agents from code. CodeSandbox SDK gives you the VMs, and you install the agents. Replit's MCP server drives Replit Agent per app, with up to 10 parallel Agent tasks on Pro. boat installs seven agents with a prompt and steer API. | 32nd cheapest of 365 $3,731 a month | 204th of 365, with limits $2,000 a month | 5th cheapest of 365 $634 a month |
| Always-on AI employees 50 machines 24/7 · 2 vCPU / 4 GB | 50th cheapest of 361 $4,811 a month | 246th of 361, with limits $2,033 a month | 6th cheapest of 361 $657 a month |
| Code interpreter bursts 100 at once · 1 vCPU / 2 GB · 200,000 sessions of 1 min a month CodeSandbox SDK fits short code runs, with VMs that hibernate with their memory and wake on the next request. Replit's Autoscale bills only while it serves and comes to $146 a month in our code-interpreter scenario against $642, but it is a web host: you set up the code runner yourself as an HTTP app. | 54th cheapest of 336 $642 a month | 225th of 336, with limits $146 a month | 189th of 336, with limits $2,000 a month |
| Sandboxes for your users 300 at once · 2 vCPU / 4 GB · 30,000 sessions of 23 min a month CodeSandbox SDK caps Scale at 250 VMs at once, and Replit does not publish its limit, so 300 user sandboxes are not confirmed on either. boat allows 300 at once on its $100 plan. | 140th of 349, with limits $1,707 a month | 229th of 349, with limits $825 a month | 3rd cheapest of 349 $209 a month |
| RL rollouts 1,000 at once · 2 vCPU / 4 GB · 1,000,000 sessions of 3 min a month CodeSandbox SDK's memory forks let many rollouts branch from the same running state. Replit's docs describe no fork. A boat fork copies the disk only. | 341st of 349, with limits $3,861 a month | 253rd of 349, with limits $4,226 a month | 39th of 349, with limits $2,000 a month |
| Agent evals 250 at once · 4 vCPU / 8 GB · 15,000 sessions of 20 min a month CodeSandbox SDK fits agent evals because your code can start and fork many VMs. Replit's docs describe no API to start many machines. boat fits too, with full VMs and start limits per plan. | 77th cheapest of 359 $1,335 a month | 266th of 359, with limits $685 a month | 10th cheapest of 359 $200 a month |
| CI runners 40 at once · 8 vCPU / 16 GB · 6,000 sessions of 12 min a month CI jobs that build containers need Docker, which CodeSandbox SDK has. Replit's Nix-based environment docs do not describe Docker. boat also adds /dev/kvm for Android and VM builds. | 96th cheapest of 374 $860 a month | 278th of 374, with limits $425 a month | 2nd cheapest of 374 $86 a month |
| 24/7 dev boxes 1 machine 24/7 · 4 vCPU / 8 GB A Replit workspace is a good place to code by hand, with SSH on paid plans. A CodeSandbox SDK VM bills every minute it runs and documents no SSH. boat gives a full VM with SSH, Docker and systemd, with no compute charge while stopped. | 316th of 371, with limits $240 a month | 181st of 371, with limits $109 a month | 19th cheapest of 371 $26 a month |
| Computer-use agents 50 at once · 4 vCPU / 8 GB · 20,000 sessions of 15 min a month Neither documents a computer-use API. On CodeSandbox SDK you add a VNC desktop to your Dockerfile; Replit's VNC pane shows a desktop to a person. boat streams a 60 fps desktop with computer tools for agents. | 54th cheapest of 367 $1,236 a month | 225th of 367, with limits $619 a month | 7th cheapest of 367 $200 a month |
| Browser-use agents 50 at once · 2 vCPU / 4 GB · 30,000 sessions of 10 min a month | 125th of 356, with limits $741 a month | 277th of 356, with limits $317 a month | 7th cheapest of 356 $200 a month |
| Scraping & bots (startup scale) 50 at once · 1 vCPU / 2 GB · 100,000 sessions of 3 min a month | 119th cheapest of 343 $871 a month | 282nd of 343, with limits $217 a month | 6th cheapest of 343 $90 a month |
| Mass scraping & bots (undetectable) 1,000 at once · 1 vCPU / 2 GB · 3,000,000 sessions of 2 min a month | 194th of 327, with limits $7,390 a month | 147th of 327, with limits $4,378 a month | 73rd of 327, with limits $1,800 a month |
| Parallel red teaming 200 at once · 2 vCPU / 4 GB · 5,000 sessions of 30 min a month | 29th cheapest of 351 $505 a month | 241st of 351, with limits $215 a month | 3rd cheapest of 351 $200 a month |
| GPU eval jobs 10 at once · 8 vCPU / 32 GB · 500 sessions of 30 min a month | Not ranked: misses a requirement | Not ranked: misses a requirement | Not ranked: misses a requirement |
| macOS agents 5 at once · 4 vCPU / 8 GB · 1,000 sessions of 20 min a month | Not ranked: misses a requirement | Not ranked: misses a requirement | Not ranked: misses a requirement |
Key features
| Feature | CodeSandbox SDK | Replit | boat |
|---|---|---|---|
| Kind of product | Sandbox API: Firecracker VMs you start from a TypeScript SDK | AI app builder with browser workspaces and app hosting | Sandboxes you create, stop and fork by API, SDK or CLI |
| Isolation | Firecracker microVM with its own kernel | Not documented for workspaces; the security page says strong logical separation between users | Full VM with its own kernel, on hosts dedicated to boat (AMD Ryzen 7 9700X and Ryzen 9 9950X); vCPUs are shared |
| Creating machines from code | SDK creates, forks and hibernates sandboxes; 1,000 new sandboxes an hour on Scale | MCP server tools to create, update and publish apps; the docs describe no API to start many machines | API, SDK and CLI to create, stop, resume and fork sandboxes, within each plan's start caps (200 a day on $20) |
| Machines at once | 250 VMs at once on Scale; more on Enterprise | Not published | 100 on $20 up to 2,000 on $2,000; the $2,000 plan includes 90 starts a minute and refuses starts above 270 a minute |
| Pause and resume | Hibernates with a memory snapshot and resumes with processes running in about 0.5 to 2 s | Workspace files persist, and Agent checkpoints save files, database and settings; a Reserved VM runs continuously | Disk saved every minute and at each stop, kept at least 14 days; memory is not saved, so a resume is like a reboot |
| Wake on request | Yes; an HTTP request to a host wakes a hibernated VM | Autoscale starts on the first request and goes idle after 15 minutes | No; you start a stopped sandbox by API |
| Docker and nested virtualization | Docker and root, built from any Dockerfile; nested virtualization not documented | System packages come from Nix; the docs do not describe Docker, systemd or /dev/kvm | Docker, root, systemd and /dev/kvm on every sandbox; you cannot boot your own Docker image |
| Shell access | Terminals, commands and files through the SDK; SSH is not documented | SSH into the workspace on paid plans | SSH, plus exec and files over the API |
| Inbound access | HTTPS host per port, public by default, signed tokens for private sandboxes; raw TCP is not documented | Published apps on HTTPS domains, custom domains and private deployments | Up to 50 HTTPS URLs, token-gated by default; raw ports on the public IPv6 address (some sandboxes also have IPv4) |
| Desktop and computer use | No built-in desktop; a headless Chromium template, and you can install a VNC desktop in your own Dockerfile | VNC pane that streams a native desktop to your browser; the docs describe no computer-use API | Built-in desktop and Chrome streamed at 1920x1080, 60 fps, with computer-use tools; no GPU |
| Coding agents | You install your own agent in the template and run it through the commands API | Replit Agent, which you can also drive from an MCP client | Seven harnesses installed, including Claude Code and Codex, with a prompt, steer and events API |
| Keeping keys out of the sandbox | Secrets passed in as environment variables; no credentials proxy documented | Replit manages keys for OpenAI, Anthropic, Google, Grok and OpenRouter models, billed to credits; other keys are secrets in the app | Optional credentials proxy ($1 a day while on, plus $1 per 1,000 requests) for supported model providers and custom HTTPS APIs |
| Databases and storage | Sandboxes only | Postgres and App Storage built into each app | No hosted databases; you run your own inside a sandbox |
| Regions | A country hint steers scheduling, but placement is not guaranteed | North America, Europe, Asia, South America and Australia for published apps on Core and up | Germany, Finland and France; you cannot pick a region; no US or Asia region |
Every feature Battleships tracks (43)
| Feature | CodeSandbox SDK | Replit | boat |
|---|---|---|---|
| Full VM (own kernel) | Yes | Yes | Yes |
| Docker inside | Yes | Not published | Yes |
| Nested virtualization | No documented way | No documented way | Yes |
| Root | Yes | Not published | Yes |
| Systemd | Not published | Not published | Yes |
| Desktop GUI | With your own setup | Yes | Yes |
| Browser | With your own setup | No documented way | Yes |
| Computer-use API | No documented way | No documented way | Yes |
| SSH | No documented way | Yes | Yes |
| Persistent disk | Yes | Yes | Yes |
| Pause/resume | Yes | No documented way | Yes |
| Automatic snapshots | Yes | Yes | Yes |
| Fork/clone | Yes | Yes | Yes |
| Live fork (no pause) | Yes | No documented way | Yes |
| Memory fork | Yes | No documented way | No documented way |
| Custom image (Docker or snapshot) | Yes | Yes | Yes |
| ≥24 h sessions | Yes | Yes | Yes |
| Idle auto-stop | Yes | Yes | No documented way |
| Wake on request | Yes | Yes | No documented way |
| HTTPS preview URLs | Yes | Yes | Yes |
| Raw TCP inbound | No documented way | Not published | Yes |
| Public IPv4 | No documented way | No documented way | No documented way |
| Egress allowlist | No documented way | No documented way | No documented way |
| Secret proxy | No documented way | Yes | Yes |
| Secret proxy for any API | No documented way | No documented way | Yes |
| Preinstalled agents | No documented way | Yes | Yes |
| Agent harness API | No documented way | Yes | Yes |
| MCP server | No documented way | Yes | No documented way |
| GPU | No documented way | No documented way | No documented way |
| Windows | No documented way | No documented way | No documented way |
| Arm64 | No documented way | Not published | No documented way |
| Self-hosting | No documented way | No documented way | No documented way |
| BYOC | No documented way | No documented way | No documented way |
| Open source | No documented way | No documented way | No documented way |
| Scoped API keys | Not published | Not published | Yes |
| Spend limits | Not published | Yes | Yes |
| Usage per sandbox (API) | No documented way | Yes | Yes |
| Audit logs | Not published | Yes | No documented way |
| SSO | Yes | Yes | No documented way |
| SOC 2 | Yes | Yes | No documented way |
| HIPAA | No documented way | No documented way | No documented way |
| EU data residency | No documented way | Yes | No documented way |
| Zero data retention | Not published | Not published | Yes |
Performance
Neither CodeSandbox SDK nor Replit is in the public CPU benchmarks we use, so we cannot say which is faster. In StarSling's run of 3 October 2026 (run 37073680030), boat did 28.06 runs/s on the V8 Web Tooling Benchmark at 4 vCPU / 8 GB, second of 15. Every boat sandbox in this run ran on an AMD Ryzen 9 9950X host. boat also places some sandboxes on AMD Ryzen 7 9700X hosts, which this run did not measure. For start time, ComputeSDK measured CodeSandbox SDK at its default size: 11.76 s median with 100 starts at once (6 November 2026) and 2.33 s one at a time in its latest published run. Replit and boat are not in ComputeSDK.
V8 Web Tooling Benchmark at a 4 vCPU / 8 GB target, runs per second, higher is better. Source: StarSling's run of 3 October 2026 (run 37073680030), median of 5 sandboxes. Providers ran on different hardware in different regions. StarSling gives a shared rank when the run cannot separate two providers. Every boat sandbox in this run ran on an AMD Ryzen 9 9950X host. boat also places some sandboxes on AMD Ryzen 7 9700X hosts, which this run did not measure. CodeSandbox SDK and Replit are not in the public benchmarks we use, so we cannot say which is faster.
Pricing in five scenarios
Each bill is priced from the providers' public price lists by the open Battleships engine, with the cheapest plan that runs the scenario. Start limits, per-start fees and plan minimums are in the price, boat's included.
| Scenario | CodeSandbox SDK, a month | Replit, a month | boat, a month |
|---|---|---|---|
| One coding agent, workdays 4 vCPU / 8 GB, 50 GB disk, 8 hours a day, 22 days a month, files kept between days | $193 · Scale | $122 · Pro $100 (annual) · with limits | $20 · $20/mo |
| One agent that never stops 4 vCPU / 8 GB, 50 GB disk, running 24/7 all month | $341 · Scale | $122 · Pro $100 (annual) · with limits | $26 · $20/mo |
| 50 agents in parallel 50 machines of 4 vCPU / 8 GB at once, 8-hour sessions, 1,100 sessions a month, Docker inside | $2,500 · Scale | $6,092 · Pro $2,500 (annual) · with limits | $317 · $20/mo × 3 seats |
| Sandboxes for your users 2 vCPU / 4 GB, 30,000 sessions of 10 minutes a month, up to 300 at once | $815 · Scale · with limits | $403 · Pro $250 (annual) · with limits | $200 · $100/mo × 2 seats |
| 1,000 short jobs at once 2 vCPU / 4 GB, 1,000 at once, 3 minutes each, 200,000 jobs a month | $1,632 · Scale · with limits | $825 · Pro $1,000 (annual) · with limits | $8,800 · $2000/mo · with limits |
CodeSandbox SDK: You buy VM credits at $0.01486 each, and each size burns a fixed number per hour: 20 for a 4 vCPU / 8 GB VM, about $0.30. Billing is per started minute while the VM runs, including idle time before it hibernates. A hibernated VM has no VM charge. Larger teams pay a $170 monthly Scale platform fee, which is not usage credit. Prices as checked on 28 September 2026; CodeSandbox's pricing pages blocked our automated check on 8 November 2026.
Replit: Replit plans (Core $20, or $18 billed annually; Pro from $100) turn into monthly credits for Agent work and hosting, then usage goes on to pay-as-you-go. A Reserved VM has a fixed monthly price per size, such as $130 for 4 vCPU / 16 GB with dedicated vCPUs, and bills busy or idle. Our bills price each agent as one Reserved VM and leave out Agent usage. That size has twice the RAM of the 4 vCPU / 8 GB we compare.
Replit: Plans cost $20, $100, $500 or $2,000 a month, and the whole plan price comes back as VM time for that month. A 4 vCPU / 8 GB sandbox costs $0.036 an hour, billed per second while it runs, busy or idle. The $20 plan includes 200 starts a day; each extra start costs $0.03 to $0.07. A stopped sandbox has no compute charge and keeps its disk for at least 14 days. Saved snapshots above the included count and the credentials proxy, while it is on, still charge.
Limits in those scenarios
CodeSandbox SDK's Scale plan allows 250 VMs at once, so 300 sandboxes for your users and 1,000 short jobs at once need Enterprise. Each VM has a 20 GB disk, and prices for more disk and for egress are not published. Replit does not fully fit any scenario: one Reserved VM is one published app that runs all month, Replit does not publish how many can run at once, and its docs do not describe Docker. boat does not fit the 1,000-job burst: the $2,000 plan includes 90 starts a minute and 2,400 a day, extra starts are paid, and starts above 270 a minute fail with 429. boat never queues a start.
| Scenario | CodeSandbox SDK | Replit | boat |
|---|---|---|---|
| One coding agent, workdays |
|
| None for this scenario |
| One agent that never stops |
|
| None for this scenario |
| 50 agents in parallel |
|
| None for this scenario |
| Sandboxes for your users |
|
| None for this scenario |
| 1,000 short jobs at once |
|
|
|
Which one should you use?
Use CodeSandbox SDK when
your code needs to start many VMs that sleep between visits and wake on the next HTTP request with their memory intact, and you accept that Together calls this SDK deprecated.
- Built to start machines from code. Good for programs that need many VMs: the SDK creates and forks sandboxes, up to 1,000 new ones an hour on Scale.
- Memory hibernation and wake on request. Useful for sandboxes that sleep between visits: an HTTP request wakes a hibernated VM with its processes still running.
- Forks that share memory. Good for branching many runs from one warm state, such as rollouts or test variants.
Use Replit when
you want Replit Agent to build a web app from a description and host it for you, with databases, hosting that sleeps when idle, or regions outside Europe.
- An app built and hosted for you. Useful if you want a working app from a description: Replit Agent builds, tests and publishes it with Postgres and storage.
- Hosting that sleeps when idle. Useful for web apps with uneven traffic: Autoscale wakes on the first request and has no compute charge while idle, plus a $2 monthly base fee.
- Regions on five continents. Useful if your users are spread out: Replit publishes apps in North America, Europe, Asia, South America and Australia.
Where boat fits
If you are choosing a place for AI agents to work for hours, look at boat. Each agent gets a full VM with SSH, Docker, systemd and /dev/kvm, with seven agents and a desktop installed. CodeSandbox SDK gives you VMs and leaves the agent to you. Replit is built around its own agent and published apps. boat has no hosted database and runs only in Germany, Finland and France.
A whole machine for the agent's tools
CodeSandbox SDK runs Docker but documents no SSH and no /dev/kvm. Replit's Nix setup does not describe Docker or systemd. Emulator builds and VM-based tests are not documented on either. boat has all of these on every sandbox.
Agents driven from outside
On CodeSandbox SDK you build the agent harness yourself and pass keys as environment variables. Replit's MCP server drives Replit Agent only. boat has a prompt, steer and events API, and an optional paid credentials proxy keeps model keys outside.
A desktop and open ports for agents
Neither documents a desktop an agent can drive: on CodeSandbox SDK you install one, and Replit's VNC pane is for a person. Both serve HTTPS only. boat streams a 60 fps desktop and takes raw ports on the sandbox's IPv6 address.
Use boat when your agents do real work for hours: installs, builds, Docker, emulators, a browser and servers other people reach, on a full VM with agents installed and an API to drive them. Compare it one to one: boat vs CodeSandbox SDK and boat vs Replit.
boat's own limits
- Plans start at $20 a month. The plan price comes back as VM time for that month. Unused time does not carry over. Source
- Starts (create, fork, resume): the $20 plan includes 12 a minute, 60 an hour and 200 a day. The $2,000 plan includes 90 a minute, 600 an hour and 2,400 a day. Extra starts cost $0.03 to $0.07. A start fails with 429 above 3 times the included starts per minute, or 6 times per hour or per day. Source
- A stopped sandbox has no compute charge. boat keeps its disk for at least 14 days. A saved snapshot keeps it longer; extra saved snapshots and the AI credentials proxy are paid. Source
- Auto-stop defaults to 1 hour. You can turn it off on a paid plan. The trial allows 2 hours at most. Source
- boat saves the disk every minute and at each stop. Memory, running programs and /tmp are not saved: a resume is like a reboot. A machine failure loses writes after the last good snapshot. Source
- Machines run in Germany, Finland and France only. Snapshot copies can be stored outside the EU. Source
- Public IPv6 and outbound IPv4. No dedicated IPv4. The IP changes at each resume. Source
- Hosts are AMD Ryzen 7 9700X and Ryzen 9 9950X. vCPUs are shared. No GPU. You cannot boot your own Docker image. Source
Questions people ask
Is CodeSandbox SDK cheaper than Replit?
It depends on the job, and Replit is marked as not a clean fit in each one, because it prices an agent as a Reserved VM app that runs all month. One coding agent on workdays is $122 on Replit against $193 on CodeSandbox SDK, and one agent all month $122 against $341. 50 agents in parallel are $6,092 on Replit against $2,500. boat comes to $20, $26 and $317 for the same three.
Which is faster, CodeSandbox SDK or Replit?
Neither is in the public CPU benchmarks we use, so we cannot say. In StarSling's run of 3 October 2026, boat did 28.06 runs/s on the V8 Web Tooling Benchmark at 4 vCPU / 8 GB, on an AMD Ryzen 9 9950X host. The Replit Reserved VM we price has dedicated vCPUs; boat vCPUs are shared.
Can CodeSandbox SDK or Replit run Docker?
CodeSandbox SDK runs Docker with root inside each VM, and you can build a sandbox from any Dockerfile. Replit sets up environments with Nix packages, and its docs do not describe running Docker; no official page says it is absent. Every boat sandbox has Docker, systemd and /dev/kvm.
Can I start many Replit machines from my own code?
Not like a sandbox API. Replit's MCP server lets an MCP client create, update and publish apps through Replit Agent, which runs up to 10 background tasks in parallel on Pro. Its docs describe no API to start many machines. CodeSandbox SDK and boat both create and fork sandboxes from code.
Is CodeSandbox SDK still maintained?
CodeSandbox is now part of Together AI. Together's docs call this SDK deprecated and name together-sandbox as its successor. The last SDK release was 2.4.2 in December 2025. Self-serve sign-up still runs through CodeSandbox.
Where do they run?
CodeSandbox SDK has no region choice, only a country hint that is not guaranteed. Replit hosts data mainly on Google Cloud in the United States, with an optional India region, and publishes apps on five continents on Core and up. boat runs on machines in Germany, Finland and France; you cannot pick a region, and snapshot copies can be stored outside the EU.
Try boat
Install the CLI, then run boat new.
7-day free trial
More comparisons
- boat vs CodeSandbox SDK
- boat vs Replit
- E2B vs CodeSandbox SDK
- Daytona vs CodeSandbox SDK
- Modal vs CodeSandbox SDK
- Vercel Sandbox vs CodeSandbox SDK
- Cloudflare Sandbox vs CodeSandbox SDK
- Blaxel vs CodeSandbox SDK
- Runloop vs CodeSandbox SDK
- Tensorlake vs CodeSandbox SDK
- Morph Cloud vs CodeSandbox SDK
- CodeSandbox SDK vs Northflank