Runpod vs Replicate

Runpod rents you GPU or CPU pods by the second, with SSH and a volume that survives stops, plus serverless workers and multi-node clusters. Replicate runs AI models behind an API: you call a public model or package your own with Cog, and it scales up with requests and down to zero.

The companies and products

Runpod

Runpod is a GPU cloud for AI work: you rent GPU or CPU pods by the second, run serverless endpoints, or book multi-node GPU clusters.

A Runpod Pod is a container started from a Docker image, with a GPU or CPU-only, SSH, a web terminal and Jupyter. Serverless endpoints run your handler in workers that scale to zero and start on a request, and Instant Clusters join 2 to 8 nodes of 8 GPUs each. It is built for teams that train, fine-tune and serve models, not as a sandbox API for agents, so this page compares the parts that overlap: CPU pods as machines for agent work.

Founded: 2022
Kind of product: GPU cloud: Pods, Serverless endpoints and multi-node Clusters
Isolation: Containers in a multi-tenant environment; containers share the host kernel
Clouds: Secure Cloud in T3/T4 data centers, and Community Cloud on vetted peer-to-peer hosts
Regions: Says 30+ regions; 24 documented data centers in the US, Canada, Europe (Czechia, Romania, Iceland, Norway, France, the Netherlands, Sweden) and Australia
Billing: Prepaid balance billed per second, default spend limit of $80 an hour that grows with account history
Compliance: Completed SOC 2 Type II and SOC 3 examinations, ISO/IEC 27001:2022 certified, HIPAA and GDPR programs

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.

Company: Founded 2019, Y Combinator Winter 2020, San Francisco; joined Cloudflare in November 2025 and continues as its own brand with the same API
Product: Model inference API: public models, fine-tunes, and custom Cog models run as autoscaling deployments
Isolation: Each model runs as a Docker container built by Cog; the docs do not name the isolation layer
Hardware: CPU (1 or 4 vCPU), T4, L40S, A100 80GB, H100 and H200, up to 8 GPUs per instance; most multi-GPU and all H200 need a committed spend contract
Run time: Each prediction times out after 30 minutes unless support raises it
Hosting: Infrastructure subprocessors (AWS, CoreWeave, GCP) listed in the United States; no region choice documented
Open source: Cog is open source and its images run anywhere Docker runs; the hosted platform is not

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. “32nd cheapest of 377” means 31 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	Runpod: price rank	Replicate: price rank	bochka: price rank
Software factory 100 at once · 4 vCPU / 8 GB · 2,200 sessions of 8 h a month A coding agent needs Docker and a lasting disk: Runpod keeps only /workspace and has no Docker inside, and a Replicate instance starts fresh from its image.	203rd of 367, with limits $2,362 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	211th of 366, with limits $2,290 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 code runs suit both scale-to-zero models: Runpod Serverless and Replicate deployments start on a request and bill per second.	227th of 345, with limits $135 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 Replicate gives your users a model endpoint that scales, Runpod does not publish how many pods can run at once, and bochka gives each user a full VM.	83rd of 361, with limits $733 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	53rd of 360, with limits $3,068 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 fits agent evals as short API runs, while Runpod is a compromise there; bochka ranks 10th with a full VM per eval.	127th of 364, with limits $614 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 Runpod CPU pods fit plain build and test jobs, but they cannot run Docker inside; Replicate has no shell for CI, and bochka runs Docker in every sandbox.	32nd cheapest of 377 $301 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 Runpod gives you SSH and keeps /workspace after a stop; Replicate has no SSH and no persistent disk; bochka keeps the whole disk.	161st of 368, with limits $93 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 desktop: Runpod pods offer a terminal, SSH and Jupyter, Replicate has none of those; bochka streams a 60fps desktop with Chromium.	189th of 369, with limits $614 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	224th of 367, with limits $307 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	225th of 346, with limits $202 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	138th of 338, with limits $4,035 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	180th of 360, with limits $161 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: Runpod ranks 13th with GPUs from RTX cards to B300 by the second, Replicate ranks 22nd with T4 to H200; bochka has no GPUs.	13th cheapest of 98 $1,001 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	Runpod	Replicate	bochka
What you get	A container pod from a Docker image, with a GPU or CPU only, plus serverless workers and clusters	A model packaged as a Cog container, called through a prediction API	A full VM with its own kernel on dedicated AMD Ryzen 9 9950X hosts
Shell access	SSH, web terminal, JupyterLab and remote VS Code or Cursor	No SSH; you call the model through the HTTP API or client libraries	SSH, root and systemd, like a normal Ubuntu server
Docker inside	No; Runpod's docs say you cannot start Docker or Docker Compose in a pod	Not documented; the model is the container	Docker, Docker Compose and /dev/kvm on every machine
Files after a stop	Only /workspace is kept, and billed while stopped; the rest of the container disk is cleared	No persistent disk; instances start fresh from the image	The whole disk is kept, saved every minute and on stop; a stopped sandbox costs nothing
Fork and snapshots	No pod snapshot or fork; templates save the image and settings	No snapshots or fork; you add replicas of the same image	Fork a running sandbox into a new one with the same disk
Run time	No documented cap on pods; a pod runs until you stop it or your balance hits $0	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	Serverless workers scale to zero and start on a request; pods do not stop on their own	Deployments scale to zero and start again on the next prediction	Sandboxes stop and resume by API; no start on an incoming request
GPUs	Dozens of Nvidia and AMD types, from RTX A4000 to B300 and MI300X, plus multi-node clusters	T4, L40S, A100 80GB, H100 and H200, up to 8 per instance; H200 needs a committed spend contract	None, CPU only
Ready-made models	None hosted for you; you bring your image or handler	Thousands of public and official models, plus fine-tuning	None; your agent calls the model APIs it needs
Reaching the machine	Up to 10 HTTP ports on a public proxy URL with no built-in auth; TCP ports on a public IP	One prediction API per model behind API token auth; no own IP or ports	HTTPS URLs for up to 50 ports, token-protected; raw TCP; dedicated IPv4 or IPv6
Desktop and computer use	No desktop	No desktop or browser product	60fps desktop with Chromium, computer use and browser control
Keeping keys out of the machine	Secrets stored encrypted and passed into the pod as environment variables	Secrets sent to the model as inputs and redacted from logs	AI credentials proxy for model providers and any HTTPS API you register
Coding agents	An MCP server and skills let your local agent manage Runpod; nothing runs an agent in a pod for you	None to run inside; Replicate hosts models your agent can call	Claude Code, Codex, OpenCode and more preinstalled, with a prompt, steer and events API
Regions	24 documented data centers in the US, Canada, Europe and Australia	Infrastructure subprocessors listed in the United States; no region choice documented	EU only: Germany, Finland, France

Every feature Battleships tracks (43)

Feature	Runpod	Replicate	bochka
Full VM (own kernel)	No	No	Yes
Docker inside	No	No	Yes
Nested virtualization	No	No	Yes
Root	Yes	Not published	Yes
Systemd	No	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	Yes	No	Yes
Automatic snapshots	No	No	Yes
Fork/clone	No	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	Yes	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	Yes	Not published	Yes
Spend limits	Yes	Yes	Yes
Usage per sandbox (API)	No	No	Yes
Audit logs	Yes	Yes	No
SSO	Yes	No	No
SOC 2	Yes	Not published	No
HIPAA	Yes	Not published	No
EU data residency	Yes	No	Yes
Zero data retention	Not published	Not published	Yes

Performance

RunpodNot measured

ReplicateNot measured

bochka26.69 runs/s

There is no comparable CPU measurement for Runpod or Replicate: neither is in the public sandbox benchmark runs we use, so we cannot say which is faster. bochka did 26.69 Node.js build-and-test runs per second at 4 vCPU / 8 GB in StarSling's run of 2026-09-29. Both Runpod and Replicate pick their hardware for GPU work, which this 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. Runpod 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	Runpod, a month	Replicate, a month	bochka, 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	$26 · Pay-as-you-go (prepaid credits) · with limits	$57 · Usage · with limits	$20 · $20/mo
One agent that never stops 4 vCPU / 8 GB, 50 GB disk, running 24/7 all month	$93 · Pay-as-you-go (prepaid credits) · with limits	$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	$1,156 · Pay-as-you-go (prepaid credits) · with limits	$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	$307 · Pay-as-you-go (prepaid credits) · 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	$614 · Pay-as-you-go (prepaid credits) · with limits	$3,580 · Usage	$2,000 · $2000/mo · with limits

Runpod: Runpod bills a running pod per second from a prepaid balance, with disk on top; a stopped pod keeps billing its /workspace volume. CPU pod rates come from Runpod's public CPU catalogue, and there are no ingress or egress fees.

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.

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

Runpod keeps only /workspace after a stop, has no Docker inside, and does not publish how many pods can run at once, so 50 agents, 300 user sandboxes and 1,000 short jobs at once are not confirmed. Replicate has no persistent disk, so it does not fit the workday agent, the always-on agent or the 50-agent fleet, and each prediction stops at 30 minutes unless support raises it. Replicate does fit the user sandboxes and the 1,000-job burst, at $1,620 and $3,580 a month. bochka fits every scenario except the burst, where its start limits queue the jobs.

Scenario	Runpod	Replicate	bochka
One coding agent, workdays	files survive a stop only in /workspace: packages installed elsewhere reset	missing persistent disk no persistent disk: files lost when it stops or sleeps disk beyond 0 GiB: price unknown egress price not published	None that bite
One agent that never stops	files survive a stop only in /workspace: packages installed elsewhere reset	missing persistent disk no persistent disk: files lost when it stops or sleeps disk beyond 0 GiB: price unknown egress price not published	None that bite
50 agents in parallel	missing Docker inside files survive a stop only in /workspace: packages installed elsewhere reset how many machines can run at once: limit not published, probably fine at 50 (the typical published limit is 50)	missing persistent disk, Docker inside no persistent disk: files lost when it stops or sleeps disk beyond 0 GiB: price unknown egress price not published	None that bite
Sandboxes for your users	not confirmed: how many machines can run at once (you need 300; the typical published limit is 50)	disk beyond 0 GiB: price unknown egress price not published	None that bite
1,000 short jobs at once	not confirmed: how many machines can run at once (you need 1,000; the typical published limit is 50)	disk beyond 0 GiB: price unknown egress price not published	max 2,400 starts/day (needs ~6,667) max 600 starts/hour (needs ~1,000) bursts above 90 starts/min are queued (a burst of 333 takes ~4 min to start)

Which one should you use?

Use Runpod when

you need GPUs you can log into, multi-node clusters, serverless workers, or machines outside the US.

A machine you can log into. If you want SSH, Jupyter and a /workspace volume that survives stops, Runpod pods give you a real shell on a GPU or CPU.
More GPU choice, up to clusters. For training or big fine-tunes, Runpod rents dozens of GPU types up to B300 and joins nodes into multi-node clusters.
Regions beyond the US. Teams that need machines in Canada, Europe or Australia can pick from Runpod's documented data centers there.

Use Replicate when

you want to call a ready-made model or serve your own with Cog as an API that scales to zero.

Models one API call away. If you just want to call an image, video, speech or language model, Replicate's catalog runs thousands without hosting anything.
Serve your own model with Cog. Package a model with open-source Cog and Replicate serves it as an API that scales with traffic, and the image runs anywhere Docker runs.
Short jobs that scale to zero. For bursts of short runs, Replicate deployments wake on the next prediction and go back to zero when traffic stops.

Where bochka fits

Runpod and Replicate are both built for running models on GPUs. If what you are really running is agents that write code, install packages, run Docker and use a browser for hours, neither gives them a full machine that keeps its disk. bochka does that on CPU, and your agent can still call Runpod or Replicate when it needs a GPU.

A full VM with Docker and /dev/kvm

A coding agent runs docker compose up, builds images and starts emulators. A Runpod pod is a container that cannot run Docker inside, and Replicate does not document it, so those steps fail and need a second machine.

The whole disk survives, and you can fork it

On Runpod, packages installed outside /workspace are gone after a stop, and a Replicate instance starts fresh from its image. bochka saves the whole disk every minute, a stopped sandbox costs nothing, and you can fork a running one.

Agents, desktop and a keys proxy built in

bochka ships Claude Code, Codex and more, driven from outside by API, plus a 60fps desktop and a credentials proxy. On Runpod and Replicate secrets reach the machine as environment variables or inputs, and you build the agent harness yourself.

Use bochka when your agents need a machine of their own for hours: Docker, root, a desktop and a whole disk that survives, with coding agents built in. Compare it one to one: bochka vs Runpod and bochka vs Replicate.

Questions people ask

Is Runpod cheaper than Replicate?

In our CPU scenarios, yes: one agent that never stops comes to $93 a month on Runpod against $237 on Replicate, and 50 agents in parallel to $1,156 against $2,851. Neither fits those fully, and bochka comes to $26 and $317 for the same two.

Which is faster, Runpod or Replicate?

We have no comparable CPU measurement for either one, so we cannot say. bochka did 26.69 runs per second on the StarSling CPU test at 4 vCPU / 8 GB.

Can Runpod or Replicate run Docker inside?

Runpod's docs say you cannot start your own Docker or use Docker Compose on a pod. Replicate builds your model into a Docker image with Cog, but running Docker inside it is not documented. A bochka sandbox runs Docker, Docker Compose and /dev/kvm.

Which has better GPUs for inference?

Both rent GPUs by the second. Runpod has more types, up to B300 and AMD MI300X, and multi-node clusters; Replicate has T4 to H200 and thousands of ready-made models. bochka has no GPUs.

How long can a job run?

A Runpod pod has no documented cap and runs until you stop it or your balance runs out. A Replicate prediction times out after 30 minutes unless support raises it. bochka has no session cap on paid plans.

Can I use either one as a sandbox for AI agents?

Runpod pods give you a shell and a /workspace volume, but no Docker inside or desktop. Replicate is a model API with no SSH or persistent disk. bochka gives each agent a full VM, and the agent can still call Runpod or Replicate for GPU work.

Try bochka

Install the CLI, then run bochka new.

Product preview