Crusoe vs Replicate
Crusoe rents whole machines: GPU and CPU VMs with SSH and a disk that survives a stop, which you set up yourself. Replicate runs AI models as containers behind a prediction API, on GPUs that scale up with requests and down to zero, with no machine to log into.
The companies and products
Crusoe
Crusoe calls itself "the AI factory company": it builds energy-first data centers and rents GPU and CPU machines on Crusoe Cloud.
Crusoe Cloud is infrastructure for training and serving models: Nvidia and AMD GPU VMs, general-purpose and storage CPU VMs, disks, object storage, managed Kubernetes and Slurm, and managed inference. You create Ubuntu VMs through a console, CLI, Terraform or API, log in over SSH and install what you need. It is not a sandbox product, so this page compares the CPU VMs where they overlap with boat.
Replicate
Replicate runs AI models with an API: tens of thousands of community and official models, plus your own models packaged with Cog.
You call a public model with one API request, fine-tune one on your data, or package your own model with Cog. Cog is Replicate's open-source tool that builds a Docker image with an HTTP prediction server. Replicate runs it on CPU or NVIDIA GPU hardware as deployments that scale up with traffic and down to zero. It is built for teams that serve models, not for agents that need a machine of their own, so this page compares the places where the two overlap.
Which use cases each one fits
For each use case, Battleships prices every sandbox provider on the same workload and sorts them from the cheapest monthly bill up. “58th cheapest of 359” means 57 of the other 358 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 | Crusoe: price rank | Replicate: price rank | boat: price rank |
|---|---|---|---|
| Software factory 100 at once · 4 vCPU / 8 GB · 2,200 sessions of 8 h a month | 246th of 365, with limits $2,996 a month | 283rd of 365, with limits $4,340 a month | 5th cheapest of 365 $634 a month |
| Always-on AI employees 50 machines 24/7 · 2 vCPU / 4 GB An assistant that lives on one machine all month needs a lasting disk. A Crusoe VM has one. A Replicate instance starts fresh from its image, so files go to storage you set up. On boat, turn auto-stop off (it defaults to 1 hour). | 271st of 361, with limits $3,000 a month | 340th of 361, with limits $11,300 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 Both are compromises for short code bursts: $268 a month on Crusoe and $295 on Replicate. Crusoe publishes no start-rate limit, and Replicate scales back to zero when traffic stops. | 131st of 336, with limits $268 a month | 255th of 336, with limits $295 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 Neither gives each user an HTTPS URL to a machine. Crusoe gives a public IPv4 you put your own proxy on. Replicate gives a model endpoint that scales with traffic. boat gives up to 50 token-gated HTTPS URLs per sandbox. | 108th of 349, with limits $951 a month | 310th of 349, with limits $3,705 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 | 77th of 349, with limits $4,055 a month | 180th of 349, with limits $17,900 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 Agent evals fit on both, at $811 a month on Crusoe and $1,440 on Replicate. boat fits at $200, with disk snapshots and forks. | 58th cheapest of 359 $811 a month | 79th cheapest of 359 $1,440 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 Crusoe fits CI runners: its VMs keep a disk, and you install Docker on them. Replicate documents no SSH, no persistent disk and no Docker inside a model container. boat fits at $86 a month and adds /dev/kvm for emulators. | 46th cheapest of 374 $395 a month | 313th of 374, with limits $864 a month | 2nd cheapest of 374 $86 a month |
| 24/7 dev boxes 1 machine 24/7 · 4 vCPU / 8 GB A dev box needs SSH and a disk that keeps your installs. A Crusoe VM has both. Replicate documents neither. boat fits at $26 a month for 730 hours. | 89th cheapest of 371 $121 a month | 312th of 371, with limits $226 a month | 19th cheapest of 371 $26 a month |
| Computer-use agents 50 at once · 4 vCPU / 8 GB · 20,000 sessions of 15 min a month Neither ships a desktop. On Crusoe you install a desktop and a VNC or RDP server on the VM. On Replicate you install a browser in your Cog image and drive it inside a prediction, which stops at 30 minutes by default. boat ships a 60 fps desktop with Chrome. | 253rd of 367, with limits $811 a month | 292nd of 367, with limits $1,296 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 | 289th of 356, with limits $405 a month | 332nd of 356, with limits $1,296 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 | 303rd of 343, with limits $403 a month | 307th of 343, with limits $439 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 | 202nd of 327, with limits $8,055 a month | 221st of 327, with limits $8,775 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 | 237th of 351, with limits $207 a month | 309th of 351, with limits $855 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 Replicate fully fits GPU eval jobs at $1,338 a month. Crusoe is a compromise at $7,803, because its H100 and H200 VMs come only as 8-GPU machines. boat has no GPU. | 89th of 102, with limits $7,803 a month | 22nd cheapest of 102 $1,338 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 requirement | Not ranked: misses a requirement | Not ranked: misses a requirement |
Key features
| Feature | Crusoe | Replicate | boat |
|---|---|---|---|
| What you get | A plain Ubuntu GPU or CPU VM you set up yourself, plus Kubernetes and Slurm | A model packaged as a Cog container and called through a prediction API | A full VM for agents with tools, a desktop and coding agents installed; you cannot boot your own Docker image |
| Shell access | SSH as a sudo user (root login deprecated) | No SSH documented; you call the model through the HTTP API or client libraries | SSH, root via sudo and systemd, like a normal Ubuntu 24.04 server |
| Docker and nested virtualization | Full VM, so you install Docker yourself (only legacy GPU images ship it); nested virtualization not documented | The model is the container; Docker inside and /dev/kvm are not documented | Docker with BuildKit and Compose, and /dev/kvm for Android emulators and VMs |
| Files after a stop | The OS disk and attached disks survive a stop; local NVMe is erased; the 128 GiB OS disk bills $10.24 a month until you delete the VM | No persistent disk documented; instances start fresh from the image, so files that must last go to storage you set up | Disk saved every minute and at each stop, kept at least 14 days, longer as a saved snapshot; memory and running programs are not saved |
| Fork and snapshots | Custom image from the boot disk, made while the VM is shut off | No disk snapshot or fork documented; a deployment adds instances of the same image | Fork a running sandbox from its last 1-minute snapshot; saved snapshots as templates |
| Run time | VMs run until you stop or delete them | Instances can stay on, but each prediction times out after 30 minutes unless support raises it | Auto-stop defaults to 1 hour; set up to 30 days, or turn it off on a paid plan for no session cap |
| Scale to zero and wake | No; you start and stop VMs yourself | Yes; deployments scale to zero and start again on the next prediction | No; you start a stopped sandbox by API, and each start counts toward the plan's start limits |
| GPUs | Nvidia GB200, B300, B200, H200, H100, A100, L40S and AMD MI355X, MI300X; on-demand, spot or reserved | T4, L40S, A100 80GB, H100 and H200, up to 8 per instance, billed per second; most multi-GPU and all H200 need a committed spend contract | None on any size; graphics run on the CPU |
| Ready-made models | Managed inference; the VMs come empty | Tens of thousands of community and official models, plus fine-tuning | None; your agent calls the model API you choose |
| Reaching the machine | Public IPv4, static or dynamic; TCP and UDP load balancers; you set up HTTPS and TLS yourself | One prediction API per model behind API token auth; no own IP or ports documented | Up to 50 HTTPS URLs, token-gated by default, and SSH; public IPv6; no static IP, and the IP changes at each resume |
| Desktop and coding agents | You set it up: install a desktop, a VNC or RDP server and each agent on the VM | No desktop product; you can install a browser in your Cog image; Replicate hosts models your agent can call | 1920x1080, 60 fps desktop with Chrome, and seven coding agents driven from outside by API |
| Keeping API keys out of the machine | You set it up: put secrets on the VM as files or variables, or run your own proxy | Secrets are sent to the model as inputs and redacted from logs afterwards | Optional paid AI credentials proxy ($1 per UTC day on, plus $1 per 1,000 requests); real keys of supported providers stay out; other secrets still enter |
| Regions | US (Texas, Virginia, Ohio, Nevada) and EU (Iceland, Norway), chosen per VM | Subprocessors listed in the United States; no region choice documented | Germany, Finland and France; you cannot pick a region; no US or Asia region |
| Self-hosting | No | Cog images run on your own infrastructure, anywhere Docker runs | No; hosted service only |
Every feature Battleships tracks (43)
| Feature | Crusoe | Replicate | 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 | Yes | Not published | Yes |
| Systemd | Not published | Not published | Yes |
| Desktop GUI | No documented way | No documented way | Yes |
| Browser | No documented way | No documented way | Yes |
| Computer-use API | No documented way | No documented way | Yes |
| SSH | Yes | No documented way | Yes |
| Persistent disk | Yes | No documented way | Yes |
| Pause/resume | Yes | No documented way | Yes |
| Automatic snapshots | No documented way | No documented way | Yes |
| Fork/clone | Yes | No documented way | Yes |
| Live fork (no pause) | No documented way | No documented way | 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 | No documented way | Yes | Yes |
| Raw TCP inbound | Yes | Not published | Yes |
| Public IPv4 | Yes | No documented way | No documented way |
| Egress allowlist | Yes | No documented way | No documented way |
| Secret proxy | No documented way | No documented way | Yes |
| Secret proxy for any API | No documented way | No documented way | Yes |
| Preinstalled agents | No documented way | No documented way | Yes |
| Agent harness API | No documented way | No documented way | Yes |
| MCP server | Yes | Yes | No documented way |
| GPU | Yes | Yes | No documented way |
| Windows | No documented way | No documented way | No documented way |
| Arm64 | Yes | No documented way | No documented way |
| Self-hosting | No documented way | Yes | No documented way |
| BYOC | No documented way | No documented way | No documented way |
| Open source | No documented way | Yes | No documented way |
| Scoped API keys | Not published | Not published | Yes |
| Spend limits | No documented way | Yes | Yes |
| Usage per sandbox (API) | Yes | No documented way | Yes |
| Audit logs | Yes | Yes | No documented way |
| SSO | Yes | No documented way | No documented way |
| SOC 2 | Yes | Not published | No documented way |
| HIPAA | Not published | Not published | No documented way |
| EU data residency | Yes | No documented way | No documented way |
| Zero data retention | Not published | Not published | Yes |
Performance
Neither Crusoe nor Replicate is in the CPU benchmark we use, so we cannot say which is faster at CPU work. In StarSling's run of 3 October 2026 (run 37073680030), boat did 28.06 runs/s on the V8 Web Tooling Benchmark at 4 vCPU / 8 GB, second of 15. Every boat sandbox in this run ran on an AMD Ryzen 9 9950X host. boat also places some sandboxes on AMD Ryzen 7 9700X hosts, which this run did not measure. Both Crusoe and Replicate are mainly chosen for GPU work, which this test does not cover.
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. Crusoe and Replicate 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 | Crusoe, a month | Replicate, 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 | $32 · On-demand | $57 · Usage · with limits | $20 · $20/mo |
| One agent that never stops 4 vCPU / 8 GB, 50 GB disk, running 24/7 all month | $121 · On-demand | $237 · Usage · 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 | $1,488 · On-demand | $2,851 · Usage · with limits | $317 · $20/mo × 3 seats |
| Sandboxes for your users 2 vCPU / 4 GB, 30,000 sessions of 10 minutes a month, up to 300 at once | $405 · On-demand | $1,620 · Usage | $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 | $811 · On-demand | $3,580 · Usage | $8,800 · $2000/mo · with limits |
Crusoe: Crusoe bills each VM per second while it runs, at $0.04 per vCPU-hour with 4 GB of memory per vCPU, so a 4 vCPU / 16 GB VM costs $0.16 an hour. Every VM also has a fixed 128 GiB OS disk at $10.24 a month, billed running or stopped. There is no plan fee and no data transfer charge today.
Replicate: Replicate bills per second by hardware tier, from prepaid credit that expires after a year. Community public models bill only for active time, and official models bill per output. Private models and deployments bill setup, idle and active time, including a few idle minutes after traffic stops, and failed or canceled runs.
Replicate: Plans cost $20, $100, $500 or $2,000 a month, and the whole plan price comes back as VM time for that month. Unused VM time expires at the end of the period. A 4 vCPU / 8 GB sandbox costs $0.036 an hour, billed per second while it runs, busy or idle. The $20 plan includes 12 starts a minute and 200 a day; extra starts cost $0.03 to $0.07. A stopped sandbox has no compute charge and keeps its disk for at least 14 days. Saved snapshots above the included count and the optional credentials proxy still charge while it is stopped.
Limits in those scenarios
Replicate does not fit the workday agent, the always-on agent or the 50-agent fleet as a work machine: it documents no persistent disk, and the fleet also needs Docker inside, which is not documented. Each prediction stops after 30 minutes unless support raises it. Replicate publishes no disk or egress price, so the scenarios count none. Crusoe fits all five scenarios, but it sells 4 vCPU only with 16 GB and sets per-project quotas you raise through support, so a large fleet may need a ticket first. boat never queues a start: on the $2,000 plan, starts above 90 a minute are paid extras and starts above 270 a minute fail with 429. So at 1,000 short jobs at once, Crusoe and Replicate fit and boat does not.
| Scenario | Crusoe | Replicate | 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 |
| None for this scenario |
| Sandboxes for your users | None for this scenario |
| None for this scenario |
| 1,000 short jobs at once | None for this scenario |
|
|
Which one should you use?
Use Crusoe when
you want VMs you log into, with GPUs from L40S to GB200 or plain CPU machines, in the US or the EU, and you are happy to set them up yourself.
- A machine with a lasting disk. If you want to log in over SSH and keep files across stops, Crusoe gives you full VMs with a persistent OS disk.
- The widest GPU range. For training or big inference, Crusoe rents Nvidia up to GB200 and AMD MI300X and MI355X, on demand, spot or reserved.
- US and EU regions with network control. Crusoe runs in the US, Iceland and Norway, with VPC firewall rules for traffic in and out and static public IPs.
Use Replicate when
you want to call or serve an AI model on GPUs as an API that scales to zero, from its catalog or packaged with Cog.
- Tens of thousands of models, one API call away. Teams that want to call an image, video, speech or language model without hosting it can start with one request.
- Scale to zero. Models that wait for traffic scale down to zero and wake on the next prediction; community public models bill only for active time.
- Cog runs anywhere. Cog, the open-source tool that packages models for Replicate, builds Docker images you can also run on your own servers.
Where boat fits
Crusoe and Replicate are both built for running models on GPUs, and boat has no GPU. boat fits agents that write code, use a browser or work for hours on a full VM with tools installed. The agent can still call Replicate or a Crusoe GPU for the model work.
Work that saves itself and forks live
A Replicate instance starts fresh, so files must go to storage you set up. A Crusoe VM keeps its disk, but a copy means shutting it off to make an image. boat saves the disk every minute and forks a running sandbox. A fork can miss the last minute of writes.
A desktop and coding agents inside
On Crusoe you install, stream and drive a desktop and agents yourself. On Replicate there is no documented shell, so a browser runs inside a prediction of up to 30 minutes by default. boat ships a 60 fps desktop and seven coding agents.
boat's own limits
boat has no GPU, no model catalog, no wake on request and no US region. Starts are capped per plan, so 1,000 short jobs at once fit on Crusoe and Replicate but not on any boat plan.
Use boat when you run agents: a full VM that saves its disk every minute, forks while running, and has a desktop, coding agents and up to 50 HTTPS URLs installed. Compare it one to one: boat vs Crusoe and boat vs Replicate.
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 Crusoe cheaper than Replicate?
In the CPU scenarios we priced, yes. One agent that never stops comes to $121 a month on Crusoe and $237 on Replicate, and 1,000 short jobs at once to $811 against $3,580. For GPU eval jobs Replicate costs less, at $1,338 a month against $7,803 on Crusoe. boat has no GPU; one agent that never stops costs $26.
Which is faster, Crusoe or Replicate?
Neither is in the CPU benchmark we use, so we cannot say. Both are mainly used for GPU work, and speed there depends on the GPU you pick. In StarSling's run of 3 October 2026, boat did 28.06 runs/s on the V8 Web Tooling Benchmark at 4 vCPU / 8 GB, on an AMD Ryzen 9 9950X host.
Can Crusoe or Replicate run Docker?
A Crusoe VM gives you a sudo user over SSH, so you install Docker yourself; only legacy GPU images ship it. Replicate builds your model into a Docker image with Cog, but running Docker inside that container is not documented. A boat sandbox runs Docker, systemd and /dev/kvm.
Can I SSH into Replicate like a Crusoe VM?
Replicate documents no SSH; you reach a model through its prediction API. Crusoe VMs and boat sandboxes both give you SSH and a disk that is still there after a stop. On boat, the disk is kept for at least 14 days after a stop, longer with a saved snapshot.
Are Crusoe or Replicate sandboxes for AI agents?
Not in the usual sense. Crusoe rents plain VMs you set up yourself, and Replicate runs models behind an API. Neither ships a desktop or a coding agent harness, and neither documents a fork of a running machine. A boat sandbox has these built in.
Does boat have GPUs?
No. boat runs CPU VMs on AMD Ryzen hosts and has no GPU on any size. If your agent needs a model on a GPU, it can call Replicate's API or a Crusoe GPU VM from inside a boat sandbox.
Try boat
Install the CLI, then run boat new.
7-day free trial