Northflank vs Agent 37
Northflank runs sandboxes as one part of a big cloud platform, with microVMs, GPUs, 17 regions and your own cloud. Agent 37 hosts one always-on agent per user in the US, on light gVisor sandboxes that sleep when idle and keep their memory, with a full VM type when you need one.
The companies and products
Northflank
Northflank is a London developer platform for running services, jobs and databases on Kubernetes, and Sandboxes is its product for running untrusted code.
A Northflank sandbox is a container image deployed as a service. CPU sandboxes run in Kata microVMs with their own kernel, and GPU sandboxes run under gVisor. You create and drive them from the JavaScript or Python SDK, the CLI or the dashboard, on Northflank's cloud or inside your own. It suits teams that want one platform for apps, databases, GPUs and sandboxes.
Agent 37
Agent 37 hosts AI agents, one isolated, always-on sandbox per user, behind one API.
You create an instance from a template (Hermes, OpenClaw, Claude Code, Codex, OpenCode, Pi, Grok or your own Docker image) and talk to it through a chat API, files, exec, SSH and HTTPS URLs. The default instance is a container enforced by gVisor. A VM type gives a full Linux machine with its own kernel for Docker, Kubernetes and Android emulators. It is built for companies that give each of their customers an agent.
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. “44th cheapest of 361” means 43 of the other 360 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 | Northflank: price rank | Agent 37: price rank | boat: price rank |
|---|---|---|---|
| Software factory 100 at once · 4 vCPU / 8 GB · 2,200 sessions of 8 h a month A coding agent installs packages all over the machine. Agent 37 keeps the disk, and its VM type adds Docker at 2.5x the compute rate. Northflank keeps attached volumes, and the rest of the disk if you snapshot before each pause. boat keeps the disk with Docker and systemd and no setup. | 186th of 365, with limits $1,413 a month | 4th cheapest of 365 $551 a month | 5th cheapest of 365 $634 a month |
| Always-on AI employees 50 machines 24/7 · 2 vCPU / 4 GB This is what Agent 37 is built for: one assistant per user that sleeps when idle and wakes on the next message. It ranks 1st here at $279 a month; boat ranks 5th at $657 and Northflank 16th at $2,551. | 44th cheapest of 361 $2,551 a month | Cheapest of 361 $279 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 Short code runs on small machines suit both: Northflank has per-second plans from 0.1 vCPU, and Agent 37's small default sandboxes rank 1st here at $22 a month. boat's start limits make a big burst a compromise. | 91st of 336, with limits $114 a month | Cheapest of 336 $22 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 Agent 37 ranks 1st here: it runs up to 1,000 awake instances once you have topped up $1,000, and sleeping ones do not count. boat ranks 2nd, with 300 at once on the $100 plan with 2 seats. Northflank does not publish its concurrency limit. | 94th of 349, with limits $810 a month | Cheapest of 349 $79 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 | 333rd of 349, with limits $1,742 a month | 9th of 349, with limits $363 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 | 139th of 359, with limits $685 a month | Cheapest of 359 $58 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 boat ranks 2nd here at $86 a month and Northflank 28th at $347. Agent 37 is a compromise, because every type stops at 4 vCPU. boat adds /dev/kvm for Android emulators and VM builds, and each job start counts against its plan's start limits. | 36th cheapest of 374 $347 a month | 227th of 374, with limits $96 a month | 2nd cheapest of 374 $86 a month |
| 24/7 dev boxes 1 machine 24/7 · 4 vCPU / 8 GB | 124th cheapest of 371 $106 a month | 3rd cheapest of 371 $12 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 Agent 37 ranks 1st here: you add a noVNC desktop through a template build or the Ubuntu VM template. boat ranks 8th and streams a 60 fps desktop with Chrome with no setup. On Northflank you install a desktop and a stream in your own image. | 189th of 367, with limits $367 a month | 3rd cheapest of 367 $120 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 | 83rd of 356, with limits $359 a month | 47th of 356, with limits $29 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 | 94th cheapest of 343 $200 a month | Cheapest of 343 $32 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 | 83rd of 327, with limits $2,031 a month | 39th of 327, with limits $648 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 Agent 37 ranks 1st for many parallel tries at $43 a month, and boat 5th at $200. Both can fork a running machine; both copy the disk, not the memory, and the fork boots fresh. | 43rd of 351, with limits $180 a month | Cheapest of 351 $43 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 Only Northflank has GPU sandboxes, from L4 to H100, H200 and B200; it ranks 5th here at $692 a month. Agent 37 and boat have no GPU. | 5th cheapest of 102 $692 a month | 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 | Northflank | Agent 37 | boat |
|---|---|---|---|
| Isolation | Kata microVM with its own kernel for CPU sandboxes; gVisor for GPU sandboxes | gVisor by default; full VM as an opt-in instance type | Every sandbox is a full VM with its own kernel, on a host dedicated to boat; vCPUs are shared |
| Docker inside | Each CPU sandbox has its own kernel, so Docker can run inside | Yes on the VM type, at 2.5x the default compute rate; the default gVisor type cannot run a Docker daemon | Docker, root and systemd in every sandbox |
| Nested virtualization | Not documented inside a sandbox | /dev/kvm on the VM type, with a ready Android template | /dev/kvm in every sandbox; Android emulators work |
| What survives a stop | Attached volumes survive (billed while paused); the rest of the disk resets unless you snapshot it before pausing and restore from it; processes reset | The disk, billed while stopped or asleep; memory is checkpointed on auto-sleep | The disk, saved every minute and on stop, kept at least 14 days; no compute charge while stopped; a resume is like a reboot |
| Sleep when idle | No idle auto-pause; bills until you pause it | Opt-in auto-sleep after a quiet period (default 15 minutes), wakes on the next request | No idle check; auto-stop is a timer (default 1 hour, up to 30 days, or off on a paid plan) |
| Snapshots and fork | Filesystem snapshots of a running sandbox, on demand or on termination (opt-in, microVM only); restore into a new sandbox; no memory | Fork any running, stopped or sleeping instance (disk copy, the fork boots fresh); nightly backups in 7 slots plus one on-demand | Disk saved every minute and on stop; fork a running sandbox from its last 1-minute snapshot; saved snapshots as templates |
| Custom images | Any container image | Your own linux/amd64 Docker image as a template | Start from boat's image or a saved snapshot; you cannot boot your own Docker or OCI image |
| Desktop and browser | No built-in desktop; Northflank's guide runs Playwright from your own image; you install a desktop and stream yourself | You set it up: noVNC desktop through a template build, or the Ubuntu VM template | Built-in Linux desktop and Chrome at 1920x1080, 60 fps, with computer tools in every sandbox |
| Coding agents | Harnesses with Claude Code, Codex, Cursor, OpenCode, Pi, OMP or Devin CLI and a lifecycle API, reached by terminal or SSH | Hermes, OpenClaw, Claude Code, Codex, Grok, OpenCode, Pi, with a chat API per instance | Seven harnesses installed, with an API to prompt, steer and stream events |
| Keeping API keys out | Header injection for listed hosts in the network policy API (documented for your own cluster); otherwise env vars or files | Managed models, search and Composio through a per-instance token, metered at cost; your own keys go in as env vars | Optional AI credentials proxy ($1 a day it is on, plus $1 per 1,000 requests) for model providers and HTTPS APIs you register |
| Inbound access | HTTP ports on a code.run domain or your own, private by default; SSH; TCP and UDP through a separate load balancer with a stable IP | HTTPS URL per port, signed URLs, custom domains; no public IP, SSH through a WebSocket tunnel | SSH over IPv4 through a shared host and port; up to 50 HTTPS URLs, token-gated by default; public IPv6; no dedicated IPv4 |
| Outbound rules | Network policies with allow lists by IP, CIDR or domain, documented for your own cluster; static egress IPs | Open internet, no allowlist documented | Open internet, no provider-side allowlist |
| Sizes | Plans from 0.1 vCPU / 0.25 GB to 20 vCPU / 40 GB | 4 vCPU maximum on every type; Performance type on dedicated cores at 4x the compute rate | 2 to 16 vCPU; 16 vCPU / 32 GB on the $100 plan and up |
| GPUs | L4, A100, H100, H200, B200, RTX PRO 6000 and more | None | None on any size |
| Regions | 17 regions in North America, South America, Europe, Asia, Australia and Africa | United States | Germany, Finland and France; no US or Asia region; you cannot pick a region |
Every feature Battleships tracks (43)
| Feature | Northflank | Agent 37 | boat |
|---|---|---|---|
| Full VM (own kernel) | Yes | No documented way | Yes |
| Docker inside | Yes | No documented way | Yes |
| Nested virtualization | No documented way | No documented way | Yes |
| Root | Not published | Yes | Yes |
| Systemd | Not published | Not published | Yes |
| Desktop GUI | No documented way | With your own setup | Yes |
| Browser | With your own setup | Yes | Yes |
| Computer-use API | No documented way | No documented way | Yes |
| SSH | Yes | Yes | Yes |
| Persistent disk | With your own setup | Yes | Yes |
| Pause/resume | Yes | Yes | Yes |
| Automatic snapshots | Yes | Yes | Yes |
| Fork/clone | Yes | Yes | Yes |
| Live fork (no pause) | Yes | Yes | Yes |
| Memory fork | No documented way | No documented way | No documented way |
| Custom image (Docker or snapshot) | Yes | Yes | Yes |
| ≥24 h sessions | Yes | Yes | Yes |
| Idle auto-stop | No documented way | Yes | No documented way |
| Wake on request | No documented way | Yes | No documented way |
| HTTPS preview URLs | Yes | Yes | Yes |
| Raw TCP inbound | Yes | Not published | Yes |
| Public IPv4 | No documented way | No documented way | No documented way |
| Egress allowlist | Yes | No documented way | No documented way |
| Secret proxy | Yes | Yes | Yes |
| Secret proxy for any API | Yes | No documented way | Yes |
| Preinstalled agents | Yes | Yes | Yes |
| Agent harness API | No documented way | Yes | Yes |
| MCP server | No documented way | No documented way | No documented way |
| GPU | Yes | No documented way | No documented way |
| Windows | No documented way | No documented way | No documented way |
| Arm64 | Yes | No documented way | No documented way |
| Self-hosting | Yes | No documented way | No documented way |
| BYOC | Yes | 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) | Yes | Yes | Yes |
| Audit logs | Yes | Not published | No documented way |
| SSO | Yes | No documented way | No documented way |
| SOC 2 | Yes | Yes | No documented way |
| HIPAA | Yes | No documented way | No documented way |
| EU data residency | Yes | No documented way | No documented way |
| Zero data retention | Not published | Not published | Yes |
Performance
Northflank Sandboxes and Agent 37 are not in the public CPU benchmark 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 that 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's latest one-at-a-time run measured a Northflank median of 83 ms at its default size; boat is not in ComputeSDK. Agent 37 says its default instances share host resources, and sells a Performance type on dedicated cores.
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. Northflank and Agent 37 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 | Northflank, a month | Agent 37, 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 | $30 · Developer Sandbox (free tier) | $6 · Wallet (any top-up) | $20 · $20/mo |
| One agent that never stops 4 vCPU / 8 GB, 50 GB disk, running 24/7 all month | $106 · Developer Sandbox (free tier) | $12 · Wallet (any top-up) | $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 | $1,290 · Developer Sandbox (free tier) | $329 · Wallet ($50 lifetime top-ups) | $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 | $359 · Developer Sandbox (free tier) · with limits | $33 · Wallet ($1,000 lifetime top-ups) | $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 | $684 · Developer Sandbox (free tier) · with limits | $73 · Wallet ($1,000 lifetime top-ups) · with limits | $8,800 · $2000/mo · with limits |
Northflank: Northflank bills per second for the plan size you pick, from $0.0038 an hour for 0.1 vCPU; 4 vCPU / 8 GB is $0.1333 an hour. There is no monthly fee and no start fee, and no idle auto-pause, so a sandbox bills until you pause it. Volumes cost $0.11657 per GiB a month, also while paused, and outbound traffic costs $0.06 per GB.
Agent 37: Agent 37 uses a prepaid wallet with no plan fee, metered per minute: $0.80 per vCPU and $0.70 per GB of RAM a month while running, and $0.09 per GB of disk a month even while stopped or asleep. Performance compute costs 4x and VM compute 2.5x the default rate. New accounts get a one-time $5 credit after adding a card.
Agent 37: Plans cost $20, $100, $500 or $2,000 a month, and a card is required, also for the 7-day trial. The whole plan price comes back as VM time for that month and does not carry over. A 4 vCPU / 8 GB sandbox costs $0.036 an hour, billed per second while it runs, with 51 GB of disk included. A stopped sandbox has no compute charge and keeps its disk for at least 14 days; extra saved snapshots and the 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
On Northflank, files outside attached volumes survive a pause only if you snapshot the filesystem first and restore from it. Northflank does not publish how many sandboxes can run at once, so 300 or 1,000 at a time is not confirmed. Agent 37 caps awake instances at 3 before your first top-up, 10 after it, 50 at $50 of total top-ups, 200 at $250 and 1,000 at $1,000; sleeping and stopped ones do not count. Every Agent 37 type stops at 4 vCPU, and its 50-agent scenario needs Docker, so it is priced on the VM type at 2.5x compute. In the 1,000-job burst, Agent 37's default type runs on shared vCPUs, which the table marks as a compromise. boat's limit is that burst too: even the $2,000 plan charges for starts above 90 a minute and refuses them above 270. boat keeps a stopped disk for at least 14 days unless you save a snapshot.
| Scenario | Northflank | Agent 37 | boat |
|---|---|---|---|
| One coding agent, workdays | None for this scenario |
| None for this scenario |
| One agent that never stops | None for this scenario |
| 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 Northflank when
you need GPUs next to the agent, regions outside the US and Europe, your own container image, SOC 2, HIPAA or ISO 27001 reports, or sandboxes inside your own cloud next to your other apps.
- GPU sandboxes. For agents that train or run models: Northflank has GPU sandboxes from L4 to H100, H200 and B200.
- Regions around the world. For users spread across the globe: Northflank runs in 17 regions in the US, Europe, Asia, Australia, South America and Africa.
- Your own cloud, one platform, audit reports. For teams that must run in their own AWS, GCP or Azure account, next to their services and databases, with compliance reports.
Use Agent 37 when
you host one mostly idle assistant per customer in the US that should sleep with its memory kept and wake on the next message, at the lowest bill in our scenarios.
- Sleeps when idle and keeps its memory. For assistants idle most of the day: an instance sleeps, keeps its memory and wakes on the next message.
- The lowest bill here. For many small, mostly idle agents: Agent 37 costs less than Northflank in every scenario we priced, and ranks 1st in several of our use cases.
- Your Dockerfile, or dedicated cores. Agent 37 builds templates from your own Docker image, and sells a Performance type on dedicated cores.
Where boat fits
If your agents do long work, like installs, builds, Docker and emulators, look at boat. Every boat sandbox is a full VM with Docker, systemd and /dev/kvm, keeps its disk with no setup, can be forked while it runs, and has the agent, a desktop and a browser inside. boat costs more than Agent 37 in most of our scenarios, runs only in Europe and limits starts per plan.
A full VM is the default
On Agent 37 Docker needs the VM type at 2.5x the compute rate, with no init system and 4 vCPU at most. Northflank does not document /dev/kvm inside a sandbox. boat has Docker, systemd and /dev/kvm in every sandbox, up to 16 vCPU.
The disk comes back with no setup
On Northflank the disk outside attached volumes survives only if you snapshot before each pause and restore from it. Agent 37 keeps the disk and can keep memory on sleep. boat saves the disk every minute with no setup; a resume is like a reboot.
Agents and keys run from outside
Northflank harnesses are reached through a terminal or SSH, and Agent 37 puts your own API keys inside the machine as env vars. boat gives a prompt, steer and events API, and an optional paid credentials proxy for HTTPS APIs you register.
Use boat when your agents work on one machine for hours and should find it as they left it, with Docker, emulators, a desktop and a browser, forks of running machines and the agent driven from your code. Compare it one to one: boat vs Northflank and boat vs Agent 37.
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 Northflank cheaper than Agent 37?
No. Agent 37 costs less in all five scenarios we priced: one agent that never stops is $12 a month on Agent 37 against $106 on Northflank, and 50 agents in parallel are $329 against $1,290. Northflank does not confirm 300 or 1,000 sandboxes at once. boat costs $26 for the always-on agent and $317 for 50 agents.
Which is faster, Northflank or Agent 37?
Neither is in the public CPU benchmark 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 a Ryzen 9 9950X host. Agent 37 sells a Performance type on dedicated cores for steady speed.
Can I run Docker on Northflank and Agent 37?
On Northflank each CPU sandbox is a Kata microVM with its own kernel, so Docker can run inside. On Agent 37 Docker works on the VM type, which bills compute at 2.5x; the default gVisor type cannot run a Docker daemon. boat runs Docker in every sandbox.
Does a sandbox keep its files when it stops?
On Northflank, files on attached volumes survive a pause, and the rest of the disk survives if you take a filesystem snapshot first and restore from it. Agent 37 keeps the disk, and on auto-sleep it also checkpoints memory. boat keeps the disk on every stop for at least 14 days with no setup; memory and running programs are not saved.
Do Northflank or Agent 37 have GPUs?
Northflank has GPU sandboxes under gVisor with cards from L4 to H100, H200, B200 and more. Agent 37 and boat have no GPU.
Where do Northflank and Agent 37 run?
Northflank runs in 17 regions across the US, Europe, Asia, Australia, South America and Africa, or in your own cloud. Agent 37 runs on dedicated servers in the United States. 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