Blaxel vs Northflank

Both run each CPU sandbox in a microVM with its own kernel. Blaxel sleeps seconds after the last connection and wakes in under 25 ms with its memory intact. Northflank is a broad platform with GPUs, 17 regions, filesystem snapshots you turn on and the option to run in your own cloud.

The companies and products

Blaxel

Blaxel is a San Francisco company that runs cloud infrastructure for AI agents. Since September 2026 it is part of Baseten.

You get sandboxes, batch jobs, MCP server hosting and a model gateway. Each sandbox is a Firecracker microVM with its own kernel. It goes to standby when nothing is connected and resumes in under 25 ms with its memory and running processes. It is built for agent products that start many sandboxes and keep them asleep most of the time.

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.

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. “15th cheapest of 365” means 14 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 workloadBlaxel: price rankNorthflank: price rankboat: price rank
Software factory
100 at once · 4 vCPU / 8 GB · 2,200 sessions of 8 h a month
Blaxel keeps files in standby. On Northflank you attach volumes, or snapshot the filesystem before a pause and restore from it. boat saves the disk with no setup and adds /dev/kvm.
15th cheapest of 365
$2,373 a month
186th of 365, with limits
$1,413 a month
5th cheapest of 365
$634 a month
Always-on AI employees
50 machines 24/7 · 2 vCPU / 4 GB
17th cheapest of 361
$2,832 a month
44th cheapest of 361
$2,551 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 runs suit both: Blaxel sleeps right away and wakes in milliseconds, and Northflank has per-second plans from 0.1 vCPU with no start fee. Each boat start counts against the plan's start limits.
18th cheapest of 336
$144 a month
91st of 336, with limits
$114 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
Blaxel's self-serve tiers stop at 200 at once, and Northflank does not publish its limit. boat runs 300 at once from the $100 plan.
105th of 349, with limits
$926 a month
94th of 349, with limits
$810 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
82nd of 349, with limits
$4,378 a month
333rd of 349, with limits
$1,742 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
140th of 359, with limits
$704 a month
139th 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
55th cheapest of 374
$431 a month
36th cheapest of 374
$347 a month
2nd cheapest of 374
$86 a month
24/7 dev boxes
1 machine 24/7 · 4 vCPU / 8 GB
Northflank gives you SSH. On Blaxel you set up SSH with Tailscale. boat has built-in SSH and saves the disk every minute with no setup.
125th cheapest of 371
$114 a month
124th cheapest of 371
$106 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
Blaxel offers a desktop image with VNC. On Northflank you install a desktop, a browser and a stream in your own image. boat streams a 60 fps desktop you can embed in your app.
17th cheapest of 367
$634 a month
189th of 367, with limits
$367 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
Blaxel gives agents a browser in the sandbox. Northflank's guide runs Playwright and its browser from your own image. boat has Chrome on a desktop in every sandbox.
16th cheapest of 356
$317 a month
83rd of 356, with limits
$359 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
19th cheapest of 343
$215 a month
94th cheapest of 343
$200 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
143rd of 327, with limits
$4,292 a month
83rd of 327, with limits
$2,031 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
Blaxel has domain, method and path rules on the provider side (public preview). Northflank network policies allow by IP, CIDR or domain, documented for your own cluster. boat gives root with nftables inside the VM.
5th cheapest of 351
$217 a month
43rd of 351, with limits
$180 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
Northflank fits, with GPU sandboxes from L4 to H100, H200 and B200. Blaxel and boat have no GPU.
Not ranked: misses a requirement5th cheapest of 102
$692 a month
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 requirementNot ranked: misses a requirementNot ranked: misses a requirement

Key features

FeatureBlaxelNorthflankboat
IsolationFirecracker microVM with its own kernelKata microVM with its own kernel for CPU; gVisor for GPUFull VM, own kernel, on hosts dedicated to boat (AMD Ryzen 7 9700X and Ryzen 9 9950X); vCPUs are shared
List price, 4 vCPU / 8 GBAbout $0.33 an hour while a connection is open; standby bills snapshot storage only$0.1333 an hour (nf-compute-400), billed per second, no plan fee$0.036 an hour while running, busy or idle, from a $20 monthly plan
Sleep when idleStandby about 5 to 15 seconds after the last connection; wakes in under 25 ms with memory and processesNo idle auto-pause; bills until you pause itAuto-stop after 1 hour by default (a timer, not idle detection); a resume is like a reboot
What survives a stopWritable files live in RAM (about half of it) and are kept in standby; one block volume at creation, or Agent Drive (private preview)Attached volumes survive (billed while paused); the rest of the disk resets unless you snapshot it before pausing and restore from itThe disk, saved every minute and at each stop; kept at least 14 days, longer as a saved snapshot
Fork and snapshotsFork with memory and running processes (private preview)Opt-in filesystem snapshots of a running or terminating sandbox, restored into a new oneAutomatic disk snapshots; fork a running sandbox from its last 1-minute snapshot
Custom imagesCustom images built from Docker imagesAny container imageStart from boat's image or a saved snapshot; you cannot boot your own Docker image
Nested virtualizationNot documented by Blaxel; upstream Firecracker does not give guests /dev/kvmNot documented inside a sandboxYes: /dev/kvm, so Android emulators and VMs run inside
Shell accessRoot and a terminal through bl connect; you set up SSH with Tailscale (documented tutorial)SSH, browser terminal or SDKBuilt-in SSH, root via sudo and real systemd
Desktop and browserDesktop image with VNC and a browserNo built-in desktop; Northflank's guide runs Playwright browser agents from your own image1920x1080 desktop at 60 fps with Chrome and computer tools; embeddable stream
Coding agentsClaude Code and Codex images you drive from their terminal or your own harnessHarnesses with Claude Code, Codex, Cursor, OpenCode, Pi, OMP or Devin CLI, reached through a terminal or SSHSeven harnesses installed, driven by a prompt, steer and events API
Keeping API keys outProxy fills secrets into outbound HTTPS requests to any host (public preview)Header injection for listed hosts in the network policy API (documented for your own cluster)Optional paid proxy ($1 a day plus $1 per 1,000 requests) for supported model providers and HTTPS APIs you register
Inbound trafficPreview URL per port, public or token-protected; no public inbound IPHTTP ports on a code.run or your own domain, private by default; SSH; public TCP and UDP through a separate load balancerSSH, up to 50 HTTPS URLs (token-gated by default), raw ports over public IPv6; the IP changes at each resume
Outbound rulesDomain allow and deny lists with method and path rules (public preview); dedicated egress IPs (free during beta)Network policies with allow lists by IP, CIDR or domain, documented for your own cluster; static egress IPsOpen internet; filter inside the VM with nftables
GPUsNoneL4, A100, H100, H200, B200, RTX PRO 6000 and moreNone
RegionsOregon, North Virginia, London, Frankfurt17 regions in North America, South America, Europe, Asia, Australia and AfricaGermany, Finland and France; you cannot pick one; no US or Asia region
Your own cloudBring your own metal on a custom quoteRun on AWS, GCP, Azure and others, or the control plane in your VPC on EnterpriseNo
Every feature Battleships tracks (43)
FeatureBlaxelNorthflankboat
Full VM (own kernel)YesYesYes
Docker insideYesYesYes
Nested virtualizationNo documented wayNo documented wayYes
RootYesNot publishedYes
SystemdNo documented wayNot publishedYes
Desktop GUIYesNo documented wayYes
BrowserYesWith your own setupYes
Computer-use APIYesNo documented wayYes
SSHWith your own setupYesYes
Persistent diskYesWith your own setupYes
Pause/resumeYesYesYes
Automatic snapshotsYesYesYes
Fork/cloneYesYesYes
Live fork (no pause)YesYesYes
Memory forkYesNo documented wayNo documented way
Custom image (Docker or snapshot)YesYesYes
≥24 h sessionsYesYesYes
Idle auto-stopYesNo documented wayNo documented way
Wake on requestYesNo documented wayNo documented way
HTTPS preview URLsYesYesYes
Raw TCP inboundNo documented wayYesYes
Public IPv4No documented wayNo documented wayNo documented way
Egress allowlistYesYesNo documented way
Secret proxyYesYesYes
Secret proxy for any APIYesYesYes
Preinstalled agentsYesYesYes
Agent harness APINo documented wayNo documented wayYes
MCP serverYesNo documented wayNo documented way
GPUNo documented wayYesNo documented way
WindowsNo documented wayNo documented wayNo documented way
Arm64No documented wayYesNo documented way
Self-hostingNo documented wayYesNo documented way
BYOCYesYesNo documented way
Open sourceNo documented wayNo documented wayNo documented way
Scoped API keysYesNot publishedYes
Spend limitsNo documented wayNot publishedYes
Usage per sandbox (API)YesYesYes
Audit logsNot publishedYesNo documented way
SSOYesYesNo documented way
SOC 2YesYesNo documented way
HIPAAYesYesNo documented way
EU data residencyYesYesNo documented way
Zero data retentionYesNot publishedYes

Performance

Blaxel18.03 runs/s
NorthflankNot measured
boat28.06 runs/s

In StarSling's run of 3 October 2026 (run 37073680030), the V8 Web Tooling Benchmark at 4 vCPU and 8 GB, median of 5 sandboxes: boat did 28.06 runs per second (second of 15) and Blaxel 18.03 (rank 7, shared). Higher is better. 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. Northflank is not in that run, so we cannot say how its CPU compares. For start time, ComputeSDK's one-at-a-time runs measured a median of 83 ms for Northflank and 251 ms for Blaxel, each at its default size; boat is 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. Blaxel measured on: Sandbox (active time, RAM-only pricing, CPU bundled). 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 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.

ScenarioBlaxel, a monthNorthflank, a monthboat, 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
$33 · Pay-as-you-go Tier 1$30 · Developer Sandbox (free tier)$20 · $20/mo
One agent that never stops
4 vCPU / 8 GB, 50 GB disk, running 24/7 all month
$120 · Pay-as-you-go Tier 1$106 · Developer Sandbox (free tier)$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
$1,489 · Pay-as-you-go Tier 1$1,290 · Developer Sandbox (free tier)$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
$396 · Pay-as-you-go Tier 2 · with limits$359 · Developer Sandbox (free tier) · 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
$876 · Pay-as-you-go Tier 2 · with limits$684 · Developer Sandbox (free tier) · with limits$8,800 · $2000/mo · with limits

Blaxel: Blaxel charges $0.0000115 per GB of allocated RAM per active second, with CPU included: 8 GB comes with 4 cores, about $0.33 an hour. Active means a connection is open. In standby you pay only snapshot storage ($0.20 per GB-month), and volumes cost $0.12 per GB-month on their full size.

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. 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.

Northflank: 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, per second while it runs, busy or idle. A stopped sandbox has no compute charge and keeps its disk for at least 14 days. Saved snapshots above the included count ($0.056 a day each) and the AI credentials proxy still charge while it is stopped. The $20 plan includes 12 starts a minute and 200 a day; extra starts cost $0.03 to $0.07.

Limits in those scenarios

Blaxel's self-serve tiers allow 10, 50 or 200 sandboxes at once, standby ones included, so 300 users or 1,000 jobs at once need a sales tier, and Tier 0 and Tier 1 delete sandboxes after up to 7 or 30 days. Northflank keeps files outside attached volumes only if you snapshot before each pause, and it does not publish how many sandboxes can run at once, so 300 or 1,000 at a time is not confirmed. boat's $2,000 plan includes 90 starts a minute, charges $0.03 to $0.07 for extra starts and refuses starts above 270 a minute, so 1,000 short jobs at once do not all start together.

ScenarioBlaxelNorthflankboat
One coding agent, workdaysNone for this scenarioNone for this scenarioNone for this scenario
One agent that never stopsNone for this scenarioNone for this scenarioNone for this scenario
50 agents in parallelNone for this scenario
  • how many machines can run at once: limit not published, probably fine at 50 (the typical published limit is 50)
None for this scenario
Sandboxes for your users
  • Pay-as-you-go Tier 2: max 200 concurrent
  • not confirmed: how many machines can run at once (you need 300; the typical published limit is 50)
None for this scenario
1,000 short jobs at once
  • Pay-as-you-go Tier 2: max 200 concurrent
  • not confirmed: how many machines can run at once (you need 1,000; the typical published limit is 50)
  • 2,400 starts/day included, paid above that (needs ~6,667)
  • 600 starts/hour included, paid above that (needs ~1,000)
  • starts above 90/min are paid, and starts above 270/min fail with 429

Which one should you use?

Use Blaxel when

your sandboxes sit idle most of the time and must wake in milliseconds with their memory, or need outbound rules per domain, method and path.

  • Sleeps in seconds, wakes in milliseconds. For sandboxes that sit idle most of the time: Blaxel sleeps seconds after the last connection and wakes in under 25 ms.
  • Pay only while connected. For bursty agents: Blaxel bills allocated RAM per active second with CPU included, and standby costs only snapshot storage.
  • Files kept with memory in standby. A sleeping Blaxel sandbox keeps its files, memory and processes, with no snapshot code. Northflank keeps files outside volumes only if you snapshot first.

Use Northflank when

you need GPUs next to the agent, regions outside Europe and the US, your own container image, or sandboxes inside your own cloud account.

  • GPU sandboxes. For agents that train or run models: Northflank has GPU sandboxes from L4 and A100 up to H100, H200 and B200.
  • Regions around the world. For users far from Europe and the US: Northflank runs in 17 regions, also in Asia, Australia, South America and Africa.
  • Your own cloud, one platform. For teams that must run in their own AWS, GCP or Azure account, next to their services, jobs and databases.

Where boat fits

Blaxel is built for sandboxes that sleep most of the time, and Northflank for teams that want one platform with GPUs and many regions. Look at boat when your agents work for hours on one machine and should find it as they left it, with no snapshot code.

A whole Linux machine

Neither Blaxel nor Northflank documents /dev/kvm inside a sandbox, so Android emulators and VMs have no documented way to run. boat gives every sandbox /dev/kvm, systemd and built-in SSH.

A disk saved with no setup

boat saves the whole disk every minute and at each stop. Blaxel keeps files in RAM unless you attach a volume, and on Northflank you snapshot before each pause. A boat resume is like a reboot, and if a machine fails you lose what was written after the last good snapshot.

Agents you drive from outside

Blaxel's agent images and Northflank's Harnesses are reached through a terminal or SSH, so driving many agents from your code means building that layer yourself. boat lets you prompt, steer and follow seven built-in agents through one API.

Use boat when your agents work on one machine for hours or days and should find it as they left it: installs, Docker, emulators, a desktop and a browser, with the agent driven from your code. Compare it one to one: boat vs Blaxel and boat vs Northflank.

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 Blaxel cheaper than Northflank?

In this page's estimates, Northflank's list price is a little lower: $30 against $33 for one agent on workdays, $106 against $120 for one agent that never stops, and $1,290 against $1,489 for 50 agents in parallel. On Northflank you snapshot before a pause to keep files outside volumes, and Blaxel bills only while a connection is open, so an agent that disconnects often pays less on Blaxel. On boat those cost $20, $26 and $317.

Which is faster, Blaxel or Northflank?

Blaxel did 18.03 runs per second on the V8 Web Tooling Benchmark at 4 vCPU and 8 GB in StarSling's run of 3 October 2026. Northflank is not in that run, so we cannot compare CPU speed. boat did 28.06 in the same run, on a Ryzen 9 9950X host.

Can Blaxel and Northflank run Docker?

Yes. Blaxel has a Docker-in-sandbox image, and Northflank CPU sandboxes have their own kernel, so Docker can run inside. Neither documents /dev/kvm for Android emulators; boat exposes it in every sandbox.

Do they keep files when a sandbox stops?

Blaxel keeps files, memory and processes in standby, and Tier 0 and Tier 1 delete sandboxes after up to 7 or 30 days. Northflank keeps attached volumes when paused, and the rest of the disk if you snapshot it first and restore from it. boat keeps the disk on every stop for at least 14 days with no extra steps.

Does Blaxel or Northflank have GPUs?

Northflank does, with GPU sandboxes from L4 to H100, H200, B200 and more under gVisor. Blaxel and boat run on CPUs only.

Where do Blaxel and Northflank run?

Blaxel runs in Oregon, North Virginia, London and Frankfurt. Northflank runs in 17 regions across North America, South America, Europe, Asia, Australia and Africa, or in your own cloud. boat runs on machines in Germany, Finland and France, and you cannot pick a region.

Try boat

Install the CLI, then run boat new.

7-day free trial

More comparisons