Verda vs Replicate
Verda rents you whole Linux machines, with Nvidia GPUs or plain CPUs, by the hour in Helsinki. Replicate runs AI models for you behind an API, billed per second, and scales them down to zero when nobody calls.
The companies and products
Verda
Verda, called DataCrunch until 2025, is a European AI cloud that rents Nvidia GPUs, CPU machines and GPU clusters.
You create a GPU or CPU instance, add a disk and an SSH key, and get a full Linux VM with root and its own public IP. Verda also sells GPU clusters with InfiniBand and serverless containers that scale to zero for inference. It is built for teams that train, fine-tune and serve AI models, not as a sandbox for agents.
Replicate
Replicate runs AI models with an API: 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, 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. “10th cheapest of 377” means 9 of the other 376 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 | Verda: 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 | 88th of 367, with limits $788 a month | 268th of 367, with limits $4,340 a month | 3rd cheapest of 367 $634 a month |
| Always-on AI employees 50 machines 24/7 · 2 vCPU / 4 GB | 133rd of 366, with limits $901 a month | 335th of 366, with limits $11,300 a month | 5th cheapest of 366 $657 a month |
| Code interpreter bursts 100 at once · 1 vCPU / 2 GB · 200,000 sessions of 1 min a month Short bursts suit Replicate's scale to zero and Verda's serverless containers, though Verda ranks 99th here and Replicate 258th. | 99th of 345, with limits $165 a month | 258th of 345, with limits $295 a month | 186th of 345, 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 Verda does not publish how many instances can run at once, and Replicate gives users a model endpoint, not a machine; boat ranks 1st. | 71st of 361, with limits $629 a month | 317th of 361, with limits $3,705 a month | Cheapest of 361 $209 a month |
| RL rollouts 1,000 at once · 2 vCPU / 4 GB · 1,000,000 sessions of 3 min a month Verda ranks 20th for RL rollouts thanks to large CPU machines, ahead of boat at 34th and Replicate at 174th. | 20th of 360, with limits $1,097 a month | 174th of 360, with limits $17,900 a month | 34th of 360, with limits $2,000 a month |
| Agent evals 250 at once · 4 vCPU / 8 GB · 15,000 sessions of 20 min a month Replicate ranks 67th and fits agent evals, Verda ranks 91st as a compromise, and boat ranks 10th. | 91st of 364, with limits $123 a month | 67th cheapest of 364 $1,440 a month | 10th cheapest of 364 $200 a month |
| CI runners 40 at once · 8 vCPU / 16 GB · 6,000 sessions of 12 min a month Verda ranks 10th as a CI runner on dedicated EPYC cores, while Replicate ranks 311th because a model container has no shell or Docker for builds. | 10th cheapest of 377 $142 a month | 311th of 377, with limits $864 a month | 4th cheapest of 377 $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, which a Verda VM has and Replicate does not. | 36th cheapest of 368 $45 a month | 306th of 368, with limits $226 a month | 18th cheapest of 368 $26 a month |
| Computer-use agents 50 at once · 4 vCPU / 8 GB · 20,000 sessions of 15 min a month Neither has a built-in desktop: Verda has a guide to install VNC yourself and Replicate has none, while boat streams a 60fps desktop. | 82nd of 369, with limits $123 a month | 263rd of 369, with limits $1,296 a month | 6th cheapest of 369 $200 a month |
| Browser-use agents 50 at once · 2 vCPU / 4 GB · 30,000 sessions of 10 min a month | 169th of 367, with limits $110 a month | 322nd of 367, with limits $1,296 a month | 7th cheapest of 367 $200 a month |
| Scraping & bots (startup scale) 50 at once · 1 vCPU / 2 GB · 100,000 sessions of 3 min a month | 184th of 346, with limits $103 a month | 272nd of 346, with limits $439 a month | 6th cheapest of 346 $90 a month |
| Mass scraping & bots (undetectable) 1,000 at once · 1 vCPU / 2 GB · 3,000,000 sessions of 2 min a month | 84th of 338, with limits $2,057 a month | 221st of 338, with limits $8,775 a month | 71st of 338, with limits $1,800 a month |
| Parallel red teaming 200 at once · 2 vCPU / 4 GB · 5,000 sessions of 30 min a month | 122nd of 360, with limits $66 a month | 306th of 360, with limits $855 a month | 2nd cheapest of 360 $200 a month |
| GPU eval jobs 10 at once · 8 vCPU / 32 GB · 500 sessions of 30 min a month Both fit GPU eval jobs: Verda ranks 14th with whole GPU machines, Replicate ranks 22nd with per-second GPUs, and boat has no GPUs. | 14th cheapest of 98 $1,039 a month | 22nd cheapest of 98 $1,338 a month | Can't run it |
| macOS agents 5 at once · 4 vCPU / 8 GB · 1,000 sessions of 20 min a month | Can't run it | Can't run it | Can't run it |
Key features
| Feature | Verda | Replicate | boat |
|---|---|---|---|
| Kind of product | GPU and CPU cloud: raw VMs, GPU clusters and serverless containers | Model inference API: public models, fine-tunes and your own Cog models | Sandboxes for AI agents: VMs with agents, desktop and hosting ready |
| Isolation | Every instance is a full VM on KVM/QEMU | Each model runs as a Docker container built by Cog; the isolation layer is not named | Every sandbox is a full VM with its own kernel |
| GPUs | Nvidia V100 up to GB300, on-demand or spot, plus clusters with InfiniBand | T4, L40S, A100 80GB, H100 and H200, up to 8 per instance; most multi-GPU and all H200 need a spend contract | None, CPU only |
| Ready-made models | None listed; you run your own software or Docker image | Thousands of public and official models, plus fine-tuning | None; you bring the model API you call |
| Biggest machine | CPU instances up to 360 vCPU / 1,440 GB | CPU at 1 or 4 vCPU, or up to 8 GPUs per instance | 16 vCPU / 32 GB on the $100 plan and up |
| Shell and reaching the machine | SSH, root and a free public IPv4; you set up TLS and the firewall yourself | No SSH, no own IP or ports; one prediction API per model behind token auth | SSH and root, HTTPS URLs for up to 50 ports behind a token, raw TCP, dedicated IPv4 or IPv6 |
| Docker and nested virtualization | Root on your own VM, so you can install Docker; nested virtualization is not documented | The model is the container; Docker inside and /dev/kvm are not documented | Docker, root, systemd and /dev/kvm on every machine |
| Files after a stop | Volumes stay, but a shut-down instance keeps billing; only Delete stops the charge | No persistent disk; instances start fresh, and API inputs and outputs are deleted after an hour by default | Disk saved every minute and on stop; a stopped sandbox costs nothing and keeps it |
| Snapshots and fork | Clone a volume once its instance is shut down or it is detached | No snapshots or fork; you add replicas of the same image | Snapshots every minute and on stop; fork a running sandbox |
| Run time | No documented maximum; a VM runs until you stop or delete it | Instances can stay on, but each prediction times out after 30 minutes unless support raises it | No session cap on paid plans; you set the time to live |
| Scale to zero and wake | Serverless containers scale to zero and start on the next request | Deployments scale to zero and start on the next prediction, with min and max instances | No wake on request; you start a stopped sandbox through the API |
| Desktop and agents | No built-in desktop, only a guide to install XFCE and TigerVNC; an MCP server to manage machines | No desktop or browser product, and no agent harness to run inside | 60fps desktop with Chromium and computer use; Claude Code, Codex and more preinstalled with a prompt, steer and events API |
| Keeping keys out of the machine | Secrets only for serverless containers; on instances you place keys yourself | Secrets are sent to the model as inputs and redacted from logs afterwards | AI credentials proxy for model providers and any HTTPS API you register |
| Regions | EU only: three data centers in Helsinki, Finland | Infrastructure subprocessors listed in the United States; no region choice documented | EU only: Germany, Finland, France |
Every feature Battleships tracks (43)
| Feature | Verda | Replicate | boat |
|---|---|---|---|
| Full VM (own kernel) | Yes | No | Yes |
| Docker inside | Yes | No | Yes |
| Nested virtualization | Not published | No | Yes |
| Root | Yes | Not published | Yes |
| Systemd | Not published | Not published | Yes |
| Desktop GUI | No | No | Yes |
| Browser | No | No | Yes |
| Computer-use API | No | No | Yes |
| SSH | Yes | No | Yes |
| Persistent disk | Yes | No | Yes |
| Pause/resume | No | No | Yes |
| Automatic snapshots | No | No | Yes |
| Fork/clone | Yes | No | Yes |
| Live fork (no pause) | No | No | Yes |
| Memory fork | No | No | No |
| Custom image (Docker or snapshot) | Yes | Yes | Yes |
| ≥24 h sessions | Yes | Yes | Yes |
| Idle auto-stop | No | Yes | No |
| Wake on request | Yes | Yes | No |
| HTTPS preview URLs | Yes | Yes | Yes |
| Raw TCP inbound | No | Not published | Yes |
| Public IPv4 | Yes | No | No |
| Egress allowlist | No | No | No |
| Secret proxy | No | No | Yes |
| Secret proxy for any API | No | No | Yes |
| Preinstalled agents | No | No | Yes |
| Agent harness API | No | No | Yes |
| MCP server | Yes | Yes | No |
| GPU | Yes | Yes | No |
| Windows | No | No | No |
| Arm64 | No | No | No |
| Self-hosting | No | Yes | No |
| BYOC | No | No | No |
| Open source | No | Yes | No |
| Scoped API keys | Not published | Not published | Yes |
| Spend limits | Yes | Yes | Yes |
| Usage per sandbox (API) | Yes | No | Yes |
| Audit logs | Yes | Yes | No |
| SSO | No | No | No |
| SOC 2 | Yes | Not published | No |
| HIPAA | No | Not published | No |
| EU data residency | Yes | No | Yes |
| Zero data retention | Not published | Not published | Yes |
Performance
There is no comparable measurement for either: neither Verda nor Replicate was part of the public StarSling CPU benchmark run, so we cannot say which is faster. boat did 26.69 Node.js build-and-test runs per second at 4 vCPU / 8 GB in that run, on dedicated AMD Ryzen 9 9950X hosts. Verda says its CPU instances use dedicated AMD EPYC cores at 3.3 GHz, and Replicate's hardware is chosen for GPU inference, which the test does not cover.
Node.js build-and-test loop at 4 vCPU / 8 GB, higher is better. Measured with the open hpc-sandbox-benchmarks suite. Verda and Replicate have no published run in this suite yet, so there is no like-for-like number to compare.
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.
| Scenario | Verda, 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 | $18 · Pay as you go | $57 · Usage · with limits | $20 · $20/mo |
| One agent that never stops 4 vCPU / 8 GB, 50 GB disk, running 24/7 all month | $45 · Pay as you go | $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 | $622 · Pay as you go | $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 | $254 · Pay as you go · with limits | $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 | $507 · Pay as you go · with limits | $3,580 · Usage | $2,000 · $2000/mo · with limits |
Verda: Verda charges a prepaid balance in 10-minute blocks and refunds the unused part when you delete early. CPU machines cost $0.012 per vCPU-hour with 4 GB of RAM each, disk is extra at $0.20/GiB-month, and a shut-down instance keeps billing until you delete it.
Replicate: Replicate bills per second by hardware tier, from prepaid credit that expires after a year. Public models bill only while they run; private models and deployments also bill setup and idle time, plus failed or canceled runs.
Replicate: Plans start at $20 a month, and the whole fee is usage credit spent per second on running sandboxes. A stopped sandbox costs nothing and keeps its disk.
Limits in those scenarios
Replicate does not fit the workday agent, the always-on agent or the 50-agent fleet, because it has no persistent disk, and the fleet also needs Docker inside, which is not documented. Verda fits those three, but it does not publish how many instances an account can run at once, so 300 sandboxes for your users or 1,000 short jobs at once are not confirmed until you get a higher quota; Replicate fits both. Verda's smallest CPU machine is 4 vCPU / 16 GB, and neither publishes an egress price. boat's $2,000 plan queues the 1,000-job burst because of its start limits.
| Scenario | Verda | Replicate | boat |
|---|---|---|---|
| One coding agent, workdays |
|
| None that bite |
| One agent that never stops |
|
| None that bite |
| 50 agents in parallel |
|
| None that bite |
| Sandboxes for your users |
|
| None that bite |
| 1,000 short jobs at once |
|
|
|
Which one should you use?
Use Verda when
you want whole machines with root: GPU VMs up to GB300, GPU clusters, very large CPU machines, all in Finland.
- Whole machines with root. For teams that want a full Linux VM with SSH, root and a public IP, on a GPU or a plain CPU.
- Bigger and newer hardware. For big training or one huge job: Verda has GPUs up to GB300, InfiniBand clusters and CPU machines up to 360 vCPU.
- Data stays in Finland. For teams that need to stay in the EU: all three Verda data centers are in Helsinki.
Use Replicate when
you want to call or serve an AI model as an API, from its catalog or packaged with Cog, with GPUs billed per second that scale to zero.
- Thousands of models, one API call away. For teams that want image, video, speech or text models without hosting them: call one with a single request.
- Per-second GPUs that scale to zero. For inference with uneven traffic: deployments scale with requests, down to zero, and bill per second.
- Cog runs anywhere. For teams that want a way out: Cog is open source and its images run anywhere Docker runs.
Where boat fits
Verda and Replicate are both built for running AI models, on rented GPU machines or behind an API. If what you are really running is agents that need a machine to work inside, boat gives each one a full VM with a desktop, coding agents and hosting ready, which costs nothing while stopped. boat has no GPUs, so model training and inference still belong on Verda or Replicate.
A disk that stays, and free stops
Agents sit idle between tasks. A shut-down Verda instance keeps billing at its full rate, and deleting it leaves a volume that still bills. Replicate has no persistent disk, so an agent's files are gone when the instance scales down.
Agents, desktop and hosting ready
A Verda VM is a clean Ubuntu machine: you install the agent, a browser, a VNC server, TLS and the firewall yourself. Replicate has no shell, desktop or open ports at all. boat ships agents, a 60fps desktop and HTTPS on 50 ports.
Docker, /dev/kvm and long runs
Coding agents run docker build and emulators. Verda does not document nested virtualization, and Replicate documents neither Docker inside nor /dev/kvm, and stops each prediction at 30 minutes. boat has both and no session cap on paid plans.
Use boat when you run agents: full VMs with coding agents, a desktop, Docker and /dev/kvm, HTTPS ports and a credentials proxy, that cost nothing while stopped. Compare it one to one: boat vs Verda and boat vs Replicate.
Questions people ask
Is Verda cheaper than Replicate?
In every scenario on this page, yes. One agent that never stops costs $45 a month on Verda and $237 on Replicate, and 50 agents in parallel cost $622 against $2,851. For 1,000 short jobs at once Verda comes to $507 and Replicate to $3,580, though only Replicate is confirmed to fit.
Which is faster, Verda or Replicate?
There is no comparable measurement. Neither was part of the StarSling CPU benchmark, where boat did 26.69 runs per second at 4 vCPU / 8 GB. For GPU work, speed depends on the GPU you pick on each.
Can I run Docker on Verda or Replicate?
On Verda you get root on a full VM, so you can install Docker yourself. On Replicate your model is built into a Docker image by Cog, but running Docker inside it is not documented. boat runs Docker, systemd and /dev/kvm on every sandbox.
Does a stopped Verda instance cost money?
Yes. Verda's docs say shut-down instances continue to charge your account, and only Delete stops the instance charge. Volumes you keep bill at $0.20/GiB-month. A stopped boat sandbox costs nothing and keeps its disk.
How long can a Replicate prediction run?
Each prediction times out after 30 minutes, and you can ask Replicate for longer. A Verda VM has no documented maximum, and a boat sandbox has no session cap on paid plans.
Where do Verda and Replicate run?
Verda runs only in three data centers in Helsinki, Finland. Replicate lists its infrastructure subprocessors (AWS, CoreWeave, GCP) in the United States and documents no region choice. boat runs in Germany, Finland and France.
Try boat
Install the CLI, then run boat new.
7-day free trial