Claude Managed Agents vs Unikraft Cloud

Claude Managed Agents is a finished agent: Anthropic runs Claude and its harness, tools run in a cloud container or on a worker you host, and you send it tasks. Unikraft Cloud is raw compute: it turns your Dockerfile into a small microVM that sleeps when idle and wakes on the next request, and you bring the agent yourself.

The companies and products

Claude Managed Agents

Claude Managed Agents is Anthropic's ready-made agent harness: Claude runs as an autonomous agent, and you talk to it through an events API.

This is an agent platform, not a sandbox you rent. You define an agent (model, prompt, tools, MCP servers, skills) and an environment, then start sessions, send messages as events and stream back what Claude does. The harness handles tool calls, prompt caching and compaction. Each session gets its own Ubuntu container in Anthropic's cloud, or runs its tools on a worker that you host. It is built for teams that want Claude to do long tasks without building their own agent loop or sandbox.

Unikraft Cloud

Unikraft Cloud is a microVM cloud from Unikraft Inc. It runs any Dockerfile-based workload as a minimal Linux microVM, with cold starts it states as under 10 ms and scale-to-zero.

Your Dockerfile becomes the VM image at build time. Each instance runs on a customized Firecracker VMM and can go into standby when no traffic arrives. It is a general compute platform for dense, bursty work such as HTTP services, browsers and agent fleets. It sells flat self-serve plans, plus Enterprise dedicated hosts, on-prem and bring your own cloud. You reach instances through services, the CLI and a sandbox plugin for commands and files. For SSH or a desktop, you add them to your image with its official guides.

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. “139th cheapest of 365” means 138 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 workloadClaude Managed Agents: price rankUnikraft Cloud: price rankboat: price rank
Software factory
100 at once · 4 vCPU / 8 GB · 2,200 sessions of 8 h a month
A Managed Agents cloud sandbox has up to 8 GB of memory and Docker in limited availability. Unikraft builds your Dockerfile into the VM image, and its docs describe no container runtime inside a running instance. boat gives the agent a full VM with Docker and systemd.
139th of 365, with limits
$964 a month
Not ranked: misses a requirement5th cheapest of 365
$634 a month
Always-on AI employees
50 machines 24/7 · 2 vCPU / 4 GB
Managed Agents sessions sleep between messages with no runtime charge, and cloud sandbox files are kept 30 days after creation. Unikraft's Team pool is 8 vCPU / 8 GiB shared by every running instance; Pro is 24 vCPU / 32 GiB for $199.
259th of 361, with limits
$2,511 a month
Not ranked: misses a requirement6th cheapest of 361
$657 a month
Code interpreter bursts
100 at once · 1 vCPU / 2 GB · 200,000 sessions of 1 min a month
Both fit short bursts with long gaps: Managed Agents bills runtime only while the agent works, and Unikraft instances go into standby after a short idle cooldown and wake on the next request. A running boat sandbox bills at full rate, busy or idle.
167th of 336, with limits
$756 a month
Not ranked: misses a requirement189th 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
A Managed Agents cloud sandbox has no documented inbound ports. Unikraft Pro holds 5,000 standby instances but runs 24 at once, so hundreds of active users need an Enterprise contract. boat gives each sandbox up to 50 HTTPS URLs, token-gated by default.
227th of 349, with limits
$823 a month
Not ranked: misses a requirement3rd 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
247th of 349, with limits
$3,978 a month
Not ranked: misses a requirement39th 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
247th of 359, with limits
$320 a month
Not ranked: misses a requirement10th 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 images or run Android emulators need Docker and /dev/kvm. The Managed Agents cloud sandbox does not document /dev/kvm; a self-hosted worker can run on a KVM host you operate. Unikraft offers nested virtualization on Enterprise. boat has both on every plan.
228th of 374, with limits
$96 a month
Not ranked: misses a requirement2nd cheapest of 374
$86 a month
24/7 dev boxes
1 machine 24/7 · 4 vCPU / 8 GB
On Managed Agents you reach a cloud sandbox through the sessions API, with no documented SSH. On Unikraft you add sshd to your image and connect through a TLS tunnel. boat takes SSH directly and saves the disk every minute and at each stop.
238th of 371, with limits
$48 a month
121st cheapest of 371
$39 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
Computer-use agents need a desktop to see and click. Managed Agents ships headless Chromium with Playwright. Unikraft documents a VNC desktop recipe that you build into your image. boat streams a built-in desktop at 1920x1080, 60 fps.
166th of 367, with limits
$288 a month
Not ranked: misses a requirement7th cheapest of 367
$200 a month
Browser-use agents
50 at once · 2 vCPU / 4 GB · 30,000 sessions of 10 min a month
181st of 356, with limits
$288 a month
Not ranked: misses a requirement7th 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
194th of 343, with limits
$390 a month
Not ranked: misses a requirement6th 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
153rd of 327, with limits
$4,798 a month
Not ranked: misses a requirement73rd 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
219th of 351, with limits
$160 a month
Not ranked: misses a requirement3rd 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 requirementNot ranked: misses a requirementNot 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

FeatureClaude Managed AgentsUnikraft Cloudboat
What it isA hosted Claude agent: Anthropic's harness runs Claude, and tools run in a cloud sandbox or on a worker you hostA microVM cloud: your Dockerfile becomes a minimal Linux VM imageA sandbox platform: each agent gets its own full Linux VM, with coding agents installed
IsolationCloud sandbox: isolated Linux container per session. Self-hosted worker: the isolation of the host you runMinimal Linux microVM per instance on a customized Firecracker VMM; full VMs on EnterpriseFull VM with its own kernel, on a physical host dedicated to boat (AMD Ryzen 7 9700X or Ryzen 9 9950X); vCPUs are shared
Machine sizeCloud sandbox: up to 8 GB memory and 10 GB disk; vCPU not published. Self-hosted worker: whatever your host hasPlan pool shared by all running instances: 2 vCPU / 4 GiB, 8 vCPU / 8 GiB or 24 vCPU / 32 GiB; larger on Enterprise4 vCPU / 8 GB by default, up to 16 vCPU / 32 GB per sandbox on the $100 plan and higher; disk fixed per size, 51 to 251 GB
Docker, systemd and /dev/kvmCloud sandbox: Docker listed as limited availability; root, systemd and /dev/kvm not documentedRuns as root; no container runtime inside documented; nested virtualization on Enterprise onlyRoot, Docker, systemd and /dev/kvm on every plan
SSHCloud sandbox: no documented SSH; you use the sessions and events APIYes, with setup: run sshd in your image and connect through a TLS service and a local tunnelBuilt in on every sandbox, no tunnel
Inbound accessCloud sandbox: no documented inbound ports. Self-hosted worker: the ports you open on your hostServices on its load balancer: HTTP on 80, HTTPS on 443, TLS-wrapped TCP on other ports, custom domainsUp to 50 HTTPS URLs, token-gated by default; raw ports over public IPv6; no dedicated IPv4
Browser and desktopHeadless Chromium with Playwright; no built-in desktop or computer-use toolNo prebuilt desktop; an official VNC/noVNC desktop recipe you build into your imageBuilt-in desktop and Chrome at 1920x1080, 60 fps, with computer tools for agents
Idle and wakeAn idle session is checkpointed, not billed, and resumes on the next messageStandby after a 1 second idle cooldown by default; traffic wakes it, with memory kept in stateful modeAuto-stop timer (default 1 hour), no idle detection; you resume a stopped sandbox by API, like a reboot
What persistsCloud sandbox files kept 30 days after creation; history and memory stores stay until you delete themVolumes, plus standby snapshots within the plan's standby limitsDisk saved every minute and at each stop, kept at least 14 days after a stop; saved snapshots have no scheduled expiry
Fork and checkpointsNo documented fork or clone of a session or sandboxCheckpoint memory and volumes of a running microVM, then start new instances from itFork a running sandbox from its last 1-minute disk snapshot; memory is not copied
AgentsAnthropic's harness with Claude models, steering, multiagent threads, memory stores and webhooksNone installed; you add the agent to your own imageSeven harnesses installed (Claude Code, Codex and more), driven from outside with a prompt, steer and events API
Keeping API keys out of the sandboxVault secrets live in the sandbox as placeholders and are swapped in at egress, per hostNetwork shield injects secrets into HTTP traffic, Enterprise only, limited accessOptional paid proxy on every plan ($1 a day while on, plus $1 per 1,000 requests) for supported AI providers; other secrets enter the sandbox
Outbound rulesUnrestricted, or limited to listed hosts plus optional package registries and MCP serversNetwork shield filters traffic, Enterprise only, limited accessOpen internet, no provider-side allowlist
RegionsUS workspace geo only; inference can be global or USDallas, Frankfurt, San Francisco, Singapore and Washington DC, more on requestGermany, Finland and France; no US or Asia region; you cannot pick one
Every feature Battleships tracks (43)
FeatureClaude Managed AgentsUnikraft Cloudboat
Full VM (own kernel)No documented wayYesYes
Docker insideNot publishedNo documented wayYes
Nested virtualizationNo documented wayYesYes
RootNot publishedYesYes
SystemdNot publishedNot publishedYes
Desktop GUINo documented wayWith your own setupYes
BrowserYesWith your own setupYes
Computer-use APINo documented wayNo documented wayYes
SSHNo documented wayWith your own setupYes
Persistent diskYesYesYes
Pause/resumeYesYesYes
Automatic snapshotsYesYesYes
Fork/cloneNo documented wayYesYes
Live fork (no pause)No documented wayYesYes
Memory forkNo documented wayYesNo documented way
Custom image (Docker or snapshot)YesYesYes
≥24 h sessionsYesYesYes
Idle auto-stopYesYesNo documented way
Wake on requestYesYesNo documented way
HTTPS preview URLsNo documented wayYesYes
Raw TCP inboundNo documented wayNot publishedYes
Public IPv4No documented wayNo documented wayNo documented way
Egress allowlistYesYesNo documented way
Secret proxyYesYesYes
Secret proxy for any APIYesYesYes
Preinstalled agentsYesNo documented wayYes
Agent harness APIYesNo 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-hostingYesYesNo documented way
BYOCYesYesNo documented way
Open sourceNo documented wayNo documented wayNo documented way
Scoped API keysNot publishedNot publishedYes
Spend limitsYesNot publishedYes
Usage per sandbox (API)YesNo documented wayYes
Audit logsNot publishedYesNo documented way
SSOYesNo documented wayNo documented way
SOC 2YesYesNo documented way
HIPAANo documented wayYesNo documented way
EU data residencyNo documented wayYesNo documented way
Zero data retentionNo documented wayNot publishedYes

Performance

Claude Managed AgentsNot measured
Unikraft CloudNot measured
boat28.06 runs/s

Claude Managed Agents and Unikraft Cloud are not in the public benchmarks we use, so we cannot say which is faster. Anthropic does not publish how many vCPUs a cloud sandbox gets. For reference, boat did 28.06 runs/s on the V8 Web Tooling Benchmark at 4 vCPU / 8 GB, second of 15 in StarSling's run of 3 October 2026 (run 37073680030), on an AMD Ryzen 9 9950X host; boat also places sandboxes on other hosts. Unikraft states cold starts under 10 ms. That is its own figure and measures start time, not CPU speed.

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. Claude Managed Agents and Unikraft Cloud 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.

ScenarioClaude Managed Agents, a monthUnikraft Cloud, 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
$13 · Claude API (pay-as-you-go, usage tiers) · with limits$39 · Hobby$20 · $20/mo
One agent that never stops
4 vCPU / 8 GB, 50 GB disk, running 24/7 all month
$53 · Claude API (pay-as-you-go, usage tiers) · with limits$39 · Hobby$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
$634 · Claude API (pay-as-you-go, usage tiers) · with limitsNot ranked: misses a requirement$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
$360 · Claude API (pay-as-you-go, usage tiers) · with limitsNot ranked: misses a requirement$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
$796 · Claude API (pay-as-you-go, usage tiers) · with limitsNot ranked: misses a requirement$8,800 · $2000/mo · with limits

Claude Managed Agents: Managed Agents bills model tokens at API rates plus $0.08 per session-hour of runtime, measured to the millisecond only while the session is running. The table shows runtime only, and the cloud sandbox size is not published, so it is not a like-for-like machine. Tokens come on top, and a self-hosted worker adds the cost of your own host.

Unikraft Cloud: Unikraft charges a flat monthly fee for a pool of running capacity: free for 2 vCPU / 4 GiB with no card, $39 for 8 vCPU / 8 GiB, $199 for 24 vCPU / 32 GiB. Standby instances do not count against the pool. It publishes no per-hour, disk or egress price, so anything bigger is an Enterprise quote.

Unikraft Cloud: Plans cost $20, $100, $500 or $2,000 a month, and the whole plan price comes back as VM time for that month. A sandbox bills per second while it runs, busy or idle: $0.036 an hour for 4 vCPU / 8 GB. A stopped sandbox has no compute charge; saved snapshots above the included count and the AI credentials proxy, if on, still bill. Extra starts above the plan's included number cost $0.03 to $0.07.

Limits in those scenarios

A Managed Agents cloud sandbox has up to 8 GB of memory and 10 GB of disk, an unpublished vCPU count and Docker in limited availability, so no scenario counts as a like-for-like fit; a self-hosted worker removes these limits on a host you run. Anthropic publishes no concurrent-session cap, but usage-tier spend caps pause all API use when reached. Unikraft's limits are shared pools: Team's 8 vCPU and 8 GiB hold one 4 vCPU / 8 GB agent, so 50 agents, sandboxes for your users and 1,000 short jobs need Enterprise. On boat, the $2,000 plan includes 90 starts a minute, extra starts cost $0.03 to $0.07, and starts above 270 a minute fail with 429, so boat does not fit 1,000 jobs at once.

ScenarioClaude Managed AgentsUnikraft Cloudboat
One coding agent, workdays
  • adjacent product, not a like-for-like sandbox
  • machine size not published
  • different kind of product: agent-platform
  • machine size per session not published
  • flat pool: the tier is billed whether used or not
  • disk beyond 0 GiB: price unknown
  • egress price not published
None for this scenario
One agent that never stops
  • adjacent product, not a like-for-like sandbox
  • machine size not published
  • different kind of product: agent-platform
  • machine size per session not published
  • flat pool: the tier is billed whether used or not
  • disk beyond 0 GiB: price unknown
  • egress price not published
None for this scenario
50 agents in parallel
  • adjacent product, not a like-for-like sandbox
  • not confirmed: Docker inside
  • machine size not published
  • different kind of product: agent-platform
  • no pool tier with 200 vCPU / 400 GiB for 50 VMs
  • no published per-resource rate
None for this scenario
Sandboxes for your users
  • adjacent product, not a like-for-like sandbox
  • machine size not published
  • different kind of product: agent-platform
  • not confirmed: how many machines can run at once (you need 300; the typical published limit is 50)
  • no pool tier with 600 vCPU / 1,200 GiB for 300 VMs
  • no published per-resource rate
None for this scenario
1,000 short jobs at once
  • adjacent product, not a like-for-like sandbox
  • machine size not published
  • different kind of product: agent-platform
  • not confirmed: how many machines can run at once (you need 1,000; the typical published limit is 50)
  • no pool tier with 2,000 vCPU / 4,000 GiB for 1,000 VMs
  • no published per-resource rate
  • 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 Claude Managed Agents when

you want Claude itself to do long tasks with no harness to build, idle time that costs no runtime, host allowlists, and the option to run tools on your own workers.

  • No agent loop to build. Teams that want Claude doing long tasks get the harness, caching, compaction, memory and webhooks ready to use.
  • Idle sessions sleep with no runtime charge. Agents that wait a long time between messages pay runtime only while working, and resume on the next message.
  • Host allowlists and your own workers. Teams with strict network rules can limit sandboxes to listed hosts, or run every tool call on their own workers.

Use Unikraft Cloud when

you run many small services or browsers that sleep most of the time and must wake at once, want a free plan or flat prices, need US or Asia regions, or want GPUs or on-prem under Enterprise.

  • Fast starts and scale-to-zero. For many small services that sleep most of the time: Unikraft states cold starts under 10 ms and wakes instances on the next request.
  • Any Dockerfile, any agent. Teams that want raw compute can run any Dockerfile, with whatever agent or model they choose.
  • Regions, compliance and your own hardware. Unikraft runs in the US, Frankfurt and Singapore, has a compliance report, and Enterprise adds GPUs, arm64, on-prem or your own cloud.

Where boat fits

If you run coding agents, look at what each one gives the agent. Managed Agents gives Claude a small cloud container behind an API, or a worker you run. Unikraft gives you a microVM built from your Dockerfile, sized for short services that sleep. boat gives each agent a whole Linux VM with Docker, SSH, public URLs and a desktop, with agents installed.

A whole machine with Docker and /dev/kvm

A Managed Agents cloud sandbox has Docker in limited availability and up to 8 GB of memory. Unikraft documents no container runtime inside, and offers nested virtualization on Enterprise. On boat, docker compose and Android emulators run on every plan.

Open and log into what the agent builds

A Managed Agents cloud sandbox has no documented inbound ports or SSH. On Unikraft you add sshd and a desktop to your image and connect through a TLS tunnel. boat gives each sandbox SSH, up to 50 HTTPS URLs and a desktop at 60 fps.

Agents installed on every plan

Managed Agents runs Anthropic's harness with Claude models, and Unikraft ships no agent. boat installs seven agents. Its optional paid proxy keeps the real keys of supported AI providers out of the sandbox on every plan; Unikraft's secret injection is Enterprise only.

Use boat when your agents need a whole machine with no image work: Docker, emulators, SSH, a desktop and browser, public URLs, a disk saved with no code, and Claude Code, Codex or other agents driven from outside. Compare it one to one: boat vs Claude Managed Agents and boat vs Unikraft Cloud.

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 Claude Managed Agents cheaper than Unikraft Cloud?

For one coding agent on workdays, Managed Agents runtime is $13 a month against $39 for the Unikraft Team pool, and model tokens come on top. For an agent that never stops, Unikraft's $39 is below Managed Agents' $53 of runtime. For 50 agents or sandboxes for your users, no self-serve Unikraft plan is big enough. boat costs $20 and $26 for the first two scenarios.

Which is faster?

We cannot say: neither is in the public benchmarks we use. Unikraft states cold starts under 10 ms and wakes instances on the next request. Managed Agents resumes an idle session on the next message, and much of a session's time is spent waiting on the model.

Can Claude Managed Agents or Unikraft Cloud run Docker?

The Managed Agents cloud sandbox lists Docker as limited availability. Unikraft builds your Dockerfile into the VM image and does not document a container runtime inside a running instance. A boat sandbox is a full Ubuntu VM with Docker, systemd and /dev/kvm.

Can I SSH into either one?

A Managed Agents cloud sandbox has no documented SSH; you reach it through the sessions and events API. On Unikraft, yes, with setup: you add sshd to your image, expose it as a TLS service and connect through a local tunnel. Every boat sandbox takes SSH directly.

Can I run my own agent on them?

Managed Agents runs Anthropic's own harness with Claude models. Unikraft runs whatever agent you put in your image, and nothing is installed. boat comes with Claude Code, Codex and five other agents that you drive from outside.

Where do they run?

Managed Agents stores data and runs code in the US workspace geo only. Unikraft runs in Dallas, Frankfurt, San Francisco, Singapore and Washington DC by default. boat runs on machines in Germany, Finland and France, with no US or Asia 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