Fly.io Machines vs Modal VM Sandboxes
Fly.io Machines are microVMs for hosting your own apps near users, with no time limit; the root file system resets by default, and lasting files go on a Volume. Modal VM Sandboxes are full Linux VMs with Docker and systemd for agents. Each one runs up to 24 hours, and a filesystem snapshot carries the work to a new sandbox.
The companies and products
Fly.io Machines
Fly.io is an app hosting platform, and Machines are the microVMs underneath it that you can also start and stop through a REST API.
A Fly Machine boots your Docker image as a Firecracker microVM on Fly's own servers, in one of 17 regions. You pick a shared or performance CPU preset and mount one Volume at a time for data that must last. Fly Proxy can stop, suspend and wake Machines as traffic comes and goes. Machines themselves are an app-hosting building block; for agent sandboxes Fly also sells Sprites (see boat vs Sprites), so this page compares only the Machines product.
Modal VM Sandboxes
Modal is a serverless cloud for AI work, and VM Sandboxes are the version of its sandboxes that run in a virtual machine with their own Linux kernel.
You create a VM Sandbox from Python, TypeScript or Go code with runtime="vm", run commands in it, and terminate it when the job is done. Modal built the runtime on the Rust-based Cloud Hypervisor, so Docker, FUSE, systemd and nested cgroups work, with the same API, images and prices as its gVisor sandboxes. It became generally available on 1 October 2026. It suits teams that want many agent machines next to Modal's inference and training.
- Funding
- Over $1.2B raised
- Generally available
- 1 October 2026; early customers launched over 20 million VMs before that
- Compliance
- SOC 2 on every plan; HIPAA on Enterprise
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. “123rd cheapest of 365” means 122 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 | Fly.io Machines: price rank | Modal VM Sandboxes: price rank | boat: price rank |
|---|---|---|---|
| Software factory 100 at once · 4 vCPU / 8 GB · 2,200 sessions of 8 h a month Modal VM is a full match here with your own setup: Docker and systemd in a real VM, a desktop from its runnable example, and filesystem snapshots to continue past 24 hours. Fly is a compromise: you build the image and keep files on a Volume or with persist_rootfs. boat ships Docker, /dev/kvm and a desktop, and saves the disk at each stop. | 123rd of 365, with limits $836 a month | 27th cheapest of 365 $633 a month | 5th cheapest of 365 $634 a month |
| Always-on AI employees 50 machines 24/7 · 2 vCPU / 4 GB On Modal you snapshot the filesystem before the 24-hour end and boot a new sandbox from it each day; running programs start again. On Fly a Machine runs with no time cap, and files last on one Volume or with opt-in persist_rootfs. boat runs with no session cap once auto-stop is off (default 1 hour) on a paid plan. | 186th of 361, with limits $816 a month | Not ranked: misses a requirement | 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 Fly fits short code runs better than boat here: per-second billing with no plan fee and cheap shared CPU for code that mostly waits. boat's start caps push it to its $2,000 plan. | 83rd of 336, with limits $94 a month | Not ranked: misses a requirement | 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 Users who come back to the same sandbox fit boat: a stopped boat sandbox has no compute charge and keeps its disk for at least 14 days; extra saved snapshots and the optional keys proxy still charge. Each start counts against the plan's start cap. | 75th of 349, with limits $600 a month | Not ranked: misses a requirement | 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 Fly and boat both rank above Modal VM for RL rollouts in the default view, though none of the three is a full fit. | 52nd of 349, with limits $2,838 a month | 163rd of 349, with limits $11,982 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 | 124th of 359, with limits $457 a month | Not ranked: misses a requirement | 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 keeps the CPU busy, which drains a shared Fly vCPU's burst balance; Fly's performance CPUs avoid that at a higher rate. Modal runs Docker builds in a real VM. boat adds /dev/kvm for emulators on every plan, and each CI job counts as a start. | 154th of 374, with limits $288 a month | 170th of 374, with limits $655 a month | 2nd cheapest of 374 $86 a month |
| 24/7 dev boxes 1 machine 24/7 · 4 vCPU / 8 GB | 160th of 371, with limits $35 a month | Not ranked: misses a requirement | 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 Fly nor Modal has a desktop built in. You set it up: Modal's runnable example installs Xvfb, x11vnc, noVNC and Chromium; on Fly you add them to your image. boat ships a 60fps desktop with Chromium and computer tools. | 140th of 367, with limits $173 a month | Not ranked: misses a requirement | 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 | 229th of 356, with limits $90 a month | Not ranked: misses a requirement | 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 | 233rd of 343, with limits $67 a month | Not ranked: misses a requirement | 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 Fly's bill is lower than boat's here ($1,232 against $1,800 a month), and it can spread scrapers across 17 regions. Both are a compromise. | 47th of 327, with limits $1,232 a month | Not ranked: misses a requirement | 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 | 168th of 351, with limits $63 a month | 27th cheapest of 351 $464 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 | Fly.io Machines | Modal VM Sandboxes | boat |
|---|---|---|---|
| Kind of product | App hosting platform; you boot your own image as a microVM. Fly sells Sprites separately for agent sandboxes | Sandbox runtime inside Modal's serverless AI cloud | Agent sandbox: a full VM per agent, ready to use |
| Isolation | Firecracker microVM with its own kernel | VM with its own Linux kernel on Cloud Hypervisor | Full VM with its own kernel, on physical hosts dedicated to boat (AMD Ryzen 7 9700X and Ryzen 9 9950X); vCPUs are shared |
| Docker and systemd | Root, and Docker you install yourself; Fly's own init by default (systemd as PID 1 is not a documented setup) | Docker, systemd and root; you start dockerd yourself | Docker, root and real systemd ready in every sandbox |
| Nested virtualization | /dev/kvm not documented | Enterprise plan only, as an experimental option | /dev/kvm on every sandbox |
| CPU benchmark (V8 Web Tooling, 4 vCPU / 8 GB, StarSling run of 3 October 2026) | Not in the public benchmarks we use; shared vCPUs guarantee 6.25% of a core plus bursts, performance vCPUs get the full core | VM runtime 15.48 runs/s, gVisor 12.60 (both rank 7 of 15, shared) | 28.06 runs/s, rank 2 of 15 (shared), on a Ryzen 9 9950X host |
| Longest session | No documented cap; a Machine runs until you stop it | 24 hours (default 5 minutes); continue in a new sandbox from a filesystem snapshot | No cap once auto-stop is off on a paid plan; auto-stop defaults to 1 hour, up to 30 days |
| Files after a stop | Root file system resets by default; opt-in persist_rootfs keeps it, but Fly staff say not to trust it with critical data. One Volume per Machine (daily snapshots), billed for its full size around the clock | No paused state. You set it up: snapshot_filesystem() before the end and boot a new sandbox from it (kept 30 days by default, or with no expiry), or keep files in a Volume | The whole disk is saved every minute and at each stop, with no code; kept at least 14 days, longer with a saved snapshot. Memory, /tmp and most caches are not saved |
| Memory kept on pause | Suspend keeps memory for Machines up to 2GB | Memory snapshots in alpha; snapshotting ends the sandbox, snapshots expire after 7 days | No; a resume is like a reboot |
| Fork | Clone copies a Machine's config; its Volume starts empty unless you restore a volume snapshot or fork the Volume; the root file system is not copied | Snapshot a running sandbox's filesystem into an image and boot any number of new sandboxes from it | Fork a running sandbox from its last 1-minute snapshot; writes from the last minute or so can be missing |
| Wake on request | Fly Proxy stops or suspends idle Machines and wakes them when a request arrives | Not documented for sandboxes; your code creates a sandbox when work arrives | You start and stop sandboxes through the API; auto-stop runs on a timer, not on idle |
| Desktop and coding agents | Not built in; you add a desktop, browser, VNC and agents to your own image | Not built in. You set it up: runnable examples install Xvfb, x11vnc, noVNC, Chromium and agent harnesses | 60fps desktop, Chromium, computer tools, and seven coding-agent harnesses driven from outside |
| Keeping keys out of the sandbox | Environment variables; you deploy Fly's open-source tokenizer proxy yourself | Environment variables; you run a credential-injecting proxy in an experimental Sidecar, from an official example | Optional AI credentials proxy ($1 a day while on, plus $1 per 1,000 requests); other secrets still enter the sandbox |
| Outbound rules | Network policies allow chosen ports and protocols | Block all, IP-range allowlist, domain allowlist (beta) | Open internet, no provider-side allowlist |
| Regions | 17 regions on six continents | US, EU and Asia-Pacific; pinning costs 1.15x to 1.75x | Germany, Finland and France; you cannot pick one; no US or Asia region |
Every feature Battleships tracks (43)
| Feature | Fly.io Machines | Modal VM Sandboxes | boat |
|---|---|---|---|
| Full VM (own kernel) | Yes | Yes | Yes |
| Docker inside | Yes | Yes | Yes |
| Nested virtualization | Not published | Yes | Yes |
| Root | Yes | Yes | Yes |
| Systemd | Not published | Yes | Yes |
| Desktop GUI | No documented way | With your own setup | Yes |
| Browser | No documented way | With your own setup | Yes |
| Computer-use API | No documented way | With your own setup | Yes |
| SSH | Yes | With your own setup | Yes |
| Persistent disk | Yes | With your own setup | Yes |
| Pause/resume | Yes | No documented way | Yes |
| Automatic snapshots | Yes | No documented way | Yes |
| Fork/clone | Yes | Yes | Yes |
| Live fork (no pause) | Yes | Not published | Yes |
| Memory fork | No documented way | Yes | 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 | No documented way | No documented way |
| HTTPS preview URLs | Yes | Yes | Yes |
| Raw TCP inbound | Yes | Yes | Yes |
| Public IPv4 | Yes | No documented way | No documented way |
| Egress allowlist | Yes | Yes | No documented way |
| Secret proxy | No documented way | With your own setup | Yes |
| Secret proxy for any API | No documented way | No documented way | Yes |
| Preinstalled agents | No documented way | With your own setup | Yes |
| Agent harness API | No documented way | No documented way | Yes |
| MCP server | Yes | No documented way | 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 | No documented way | 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 | Yes | No documented way | No documented way |
| Scoped API keys | Not published | Not published | Yes |
| Spend limits | No documented way | Not published | Yes |
| Usage per sandbox (API) | No documented way | Not published | Yes |
| Audit logs | Not published | Yes | No documented way |
| SSO | No documented way | Yes | No documented way |
| SOC 2 | Yes | Yes | No documented way |
| HIPAA | Yes | Yes | No documented way |
| EU data residency | Yes | Yes | No documented way |
| Zero data retention | Not published | Not published | Yes |
Performance
In StarSling's run of 3 October 2026 (run 37073680030), on the V8 Web Tooling Benchmark at 4 vCPU / 8 GB, boat did 28.06 runs/s (second of 15), Modal's VM runtime 15.48 and Modal gVisor 12.60. StarSling ran Modal with cpu=4, and the sandbox saw 4 CPUs (PR #121); sized that way, Modal bills 4 cores, $0.76/h. 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. Providers ran on different hardware in different regions. Fly.io Machines are not in the public benchmarks we use, so we cannot say how Fly compares; its shared vCPUs guarantee 6.25% of a core with a burst balance, and its performance vCPUs get the full core at a higher rate.
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. Modal VM Sandboxes measured on: VM Sandbox, request = size. 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. Fly.io Machines is 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 | Fly.io Machines, a month | Modal VM Sandboxes, 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 | $15 · Pay As You Go · with limits | $0 · Starter | $20 · $20/mo |
| One agent that never stops 4 vCPU / 8 GB, 50 GB disk, running 24/7 all month | $36 · Pay As You Go · with limits | $109 · Starter · 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 | $487 · Pay As You Go · with limits | $1,480 · Starter | $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 | $110 · Pay As You Go · with limits | $579 · Team | $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 | $232 · Pay As You Go · with limits | $1,911 · Team · with limits | $8,800 · $2000/mo · with limits |
Fly.io Machines: Fly bills each Machine per second while it runs, by CPU preset and memory, with no plan fee. Stopped Machines pay for their root file system, and Volumes are billed for their full size all month. The table uses Fly's lowest rate: shared vCPUs bought in 1-year reservation blocks for 40% off.
Modal VM Sandboxes: Modal bills per second for the larger of the CPU and memory you request and what you use, until the sandbox ends. It prices CPU per physical core, and one core equals two vCPUs, so 4 vCPU / 8 GiB is $0.476 an hour at full request. Starter is $0 with $30 of monthly credit; Team is $250 a month with $100 of credit. Pinning a region multiplies the price by 1.15x to 1.75x. The table uses a minimal request that bursts, which needs spare host capacity.
Modal VM Sandboxes: 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. A stopped sandbox has no compute charge; extra saved snapshots and the optional AI credentials proxy still charge. The $20 plan includes 200 starts a day; extra starts cost $0.03 to $0.07 each.
Limits in those scenarios
Fly is a compromise in every scenario: its lowest figures need a 1-year commitment and use shared vCPUs that guarantee 6.25% of each core, so long builds get throttled, and files outside the one Volume reset unless you turn on persist_rootfs. A Modal VM Sandbox runs 24 hours at most, so the always-on scenario needs a snapshot and a new sandbox each day; its 1,000-job burst runs on shared burst CPU that is not guaranteed. On boat the limit that bites is starts: even the $2,000 plan includes 90 starts a minute, and starts above 270 a minute fail with 429, so 1,000 jobs at once do not fit.
| Scenario | Fly.io Machines | Modal VM Sandboxes | boat |
|---|---|---|---|
| One coding agent, workdays |
| None for this scenario | None for this scenario |
| One agent that never stops |
|
| None for this scenario |
| 50 agents in parallel |
| None for this scenario | None for this scenario |
| Sandboxes for your users |
| None for this scenario | None for this scenario |
| 1,000 short jobs at once |
|
|
|
Which one should you use?
Use Fly.io Machines when
you host web apps or light services that must run near users worldwide, sleep between requests and wake on traffic.
- 17 regions worldwide. Useful if users sit far from Europe and the US: Fly runs Machines in 17 regions, from Sydney to Johannesburg.
- No time limit and wake on request. Useful for services that sleep between requests: a Machine runs as long as you want, and Fly Proxy wakes it on traffic.
- A full app platform around it. Useful if you host whole apps: private networking, Anycast IPs, custom domains and port-level network policies come with each Machine.
Use Modal VM Sandboxes when
you run huge bursts of short agent jobs that need a real kernel and finish within a day, need strict outbound rules, or sit next to models you already run on Modal.
- A real VM made for agents. Useful when agents need Docker, FUSE and systemd: each sandbox gets its own Linux kernel and systemd, and filesystem snapshots carry work to the next sandbox.
- Very large bursts. Useful for huge waves of short jobs: Modal runs 5,000 containers at once on Team, more on Enterprise, with no per-start fee.
- Strict outbound rules and one AI platform. Useful if you serve models on Modal already: sandboxes can be limited to chosen domains and share one bill with inference.
Where boat fits
If you run coding agents that work for hours or days in a machine they come back to, boat ships that setup. Fly is an app host whose root file system resets by default, and a Modal VM runs up to 24 hours, so your code snapshots and restores it. boat gives every agent a full VM that saves its disk with no code, with a desktop and coding agents ready. boat's limits: start caps per plan, no US or Asia region, and a resume is like a reboot.
A disk saved with no code
An agent leaves tools, repos and caches all over its disk. On Fly you keep them on one Volume, billed even while stopped, or turn on persist_rootfs. On Modal you snapshot the filesystem before the 24-hour end and boot a new sandbox from it. boat saves the disk every minute and at each stop; if a machine fails, you lose what was written after the last good snapshot.
Docker, systemd and /dev/kvm ready
Fly does not document systemd as PID 1 or /dev/kvm in a Machine. Modal has systemd, and offers nested virtualization on Enterprise only, as an experimental option. On boat, Docker builds, emulators and nested VMs work on every plan from the first boot.
Desktop and agents built in
On Fly and Modal you install the desktop, browser and agent harness yourself; Modal has runnable examples for that. Keys sit in environment variables unless you run your own proxy. boat ships a 60fps desktop, seven coding-agent harnesses and an optional paid AI credentials proxy.
Use boat when your agents work for hours or days in a machine they come back to: installs, builds, tests, Docker and emulators on a VM that saves its disk with no code, with a desktop and coding agents ready. Compare it one to one: boat vs Fly.io Machines and boat vs Modal VM Sandboxes.
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 Fly.io Machines cheaper than Modal VM Sandboxes?
At Fly's lowest rate, shared vCPUs on a 1-year reservation, mostly yes: one always-on agent is $36 a month on Fly against $109 on Modal, 50 parallel agents are $487 against $1,480, and 1,000 short jobs at once are $232 against $1,911. For one agent on workdays, Modal is $0 inside Starter's $30 credit and Fly is $15. Fly's figures assume work that stays within a shared vCPU's 6.25% baseline, and Modal's assume a minimal request that bursts.
Which is faster, Fly.io Machines or Modal VM Sandboxes?
Fly.io Machines are not in the public benchmarks we use, so we cannot say. In StarSling's run of 3 October 2026, Modal's VM runtime did 15.48 V8 Web Tooling runs/s at 4 vCPU / 8 GB and boat did 28.06, on a Ryzen 9 9950X host. Fly's speed depends on whether you pick shared or performance vCPUs.
Can Fly.io Machines and Modal VM Sandboxes run Docker?
Yes. On Fly you install Docker yourself inside the microVM, which uses Fly's own init by default. Modal's VM runtime supports Docker and systemd, and you start dockerd yourself. boat has Docker and systemd ready in every sandbox.
How long can each one run?
A Fly Machine has no documented time cap and runs until you stop it. A Modal VM Sandbox defaults to 5 minutes and lasts at most 24 hours; for longer work you snapshot the filesystem and start a new sandbox from it. boat has no session cap once auto-stop is off on a paid plan (the default is 1 hour).
Do they keep my files when they stop?
Fly keeps files on the one attached Volume; the root file system resets unless you turn on persist_rootfs, which Fly says not to trust with critical data. Modal keeps files through a filesystem snapshot that a new sandbox boots from, or a Volume. boat saves the whole disk every minute and at each stop, with no code, and keeps it at least 14 days, longer with a saved snapshot.
Can I run Android emulators or nested VMs?
Fly does not document /dev/kvm inside a Machine, so nested VMs are not a documented path. Modal offers nested virtualization on the Enterprise plan only, as an experimental option. boat exposes /dev/kvm on every sandbox.
Try boat
Install the CLI, then run boat new.
7-day free trial
More comparisons
- boat vs Fly.io Machines
- boat vs Modal VM Sandboxes
- E2B vs Fly.io Machines
- Daytona vs Fly.io Machines
- Modal vs Fly.io Machines
- Vercel Sandbox vs Fly.io Machines
- Cloudflare Sandbox vs Fly.io Machines
- Blaxel vs Fly.io Machines
- Runloop vs Fly.io Machines
- Tensorlake vs Fly.io Machines
- Morph Cloud vs Fly.io Machines
- CodeSandbox SDK vs Fly.io Machines