Steel vs Hyperbrowser
Both sell cloud Chrome sessions for AI agents, with proxies and CAPTCHA solving. Steel is a focused browser API you can also run yourself as open source, while Hyperbrowser adds Firecracker sandboxes for code, fetch and search APIs, longer browser sessions and regions in the US and Europe.
The companies and products
Steel
Steel is an API that runs fleets of cloud browsers for AI agents: you create a session, connect Puppeteer, Playwright or Selenium, and release it when done.
Each Steel session is a headful Chromium in a VM in Steel's cloud, driven over CDP or through a computer-use input API, with a live viewer you can embed. Around it Steel sells managed proxies, CAPTCHA solving, saved browser profiles, a credentials vault that signs in for the agent, a files API and readable traces. It is a different kind of product from boat: a managed browser, not a general sandbox, so this page compares where the two overlap, which is agents that use a browser.
Hyperbrowser
Hyperbrowser calls itself the AI agents' gateway to the internet: cloud browsers, sandboxes, web data APIs and browser agents behind one API and one credit balance.
Its main product is a cloud Chrome session you drive with Playwright or Puppeteer, with stealth mode, residential proxies, CAPTCHA solving, saved profiles and recordings. Next to it Hyperbrowser sells Sandboxes, Firecracker VMs on Ubuntu 24.04 that live up to 24 hours, with memory snapshots, volumes and outbound rules, plus Fetch, Scrape and Search APIs and hosted browser agents. It is a browser and web data platform more than a general sandbox, so this page compares where it overlaps with boat: agents that browse, and agents that run code.
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. “25th cheapest of 367” means 24 of the other 366 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 | Steel: price rank | Hyperbrowser: price rank | boat: price rank |
|---|---|---|---|
| Software factory 100 at once · 4 vCPU / 8 GB · 2,200 sessions of 8 h a month A Steel session has no shell, and a Hyperbrowser sandbox has a shell but no documented Docker and loses its disk on stop, so builds fit boat better. | 124th of 367, with limits $1,115 a month | 136th of 367, with limits $1,206 a month | 3rd cheapest of 367 $634 a month |
| Always-on AI employees 50 machines 24/7 · 2 vCPU / 4 GB An assistant that works all month hits Steel's 1 hour session cap on Scale or Hyperbrowser's 24 hour sandbox lifetime; boat has no session cap on paid plans. | 238th of 366, with limits $3,070 a month | 255th of 366, with limits $3,650 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 | 264th of 345, with limits $333 a month | 244th of 345, with limits $246 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 300 sessions at once is more than the 100 that Steel and Hyperbrowser allow on Scale, and each user gets a browser rather than a machine. | 244th of 361, with limits $990 a month | 245th of 361, with limits $1,029 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 | 257th of 360, with limits $4,150 a month | 262nd of 360, with limits $4,972 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 | 348th of 364, with limits $400 a month | 253rd of 364, with limits $400 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 | 239th of 377, with limits $120 a month | 238th of 377, with limits $120 a month | 4th cheapest of 377 $86 a month |
| 24/7 dev boxes 1 machine 24/7 · 4 vCPU / 8 GB | 241st of 368, with limits $73 a month | 240th of 368, with limits $73 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 Steel has a computer-use input API and Hyperbrowser runs computer-use models, both inside the browser; boat streams a 60fps desktop where the agent can use any Linux app. | 148th of 369, with limits $367 a month | 144th of 369, with limits $360 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 Both fit: each hands your agent a cloud Chrome over CDP with proxies and CAPTCHA solving, and boat gives it Chromium on a full VM where it can also run code. | 25th cheapest of 367 $470 a month | 20th cheapest of 367 $360 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 All three fit at startup scale; Hyperbrowser also sells fetch and scrape APIs per page, Steel sells browser tools per call, and boat runs the scraper and its database on one machine. | 56th cheapest of 346 $540 a month | 32nd cheapest of 346 $366 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 Steel and Hyperbrowser both fit mass scraping better than boat, with residential proxies, CAPTCHA solving and stealth modes, while boat uses datacenter IPs and solves no CAPTCHAs. | 9th of 338, with limits $7,950 a month | 11th of 338, with limits $9,750 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 Hyperbrowser fits many parallel attempts better than Steel thanks to sandboxes started from memory snapshots, and boat adds a full VM you can fork while it runs. | 220th of 360, with limits $242 a month | 33rd of 360, with limits $356 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 | Can't run it | Can't run it | 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 | Steel | Hyperbrowser | boat |
|---|---|---|---|
| Kind of product | A managed Chromium browser per session, driven over CDP | Cloud browser sessions first, plus Firecracker sandboxes and web data APIs | A full Linux VM per sandbox, with a desktop and Chromium inside |
| Running code | No shell; you reach the browser through CDP and the REST API | Firecracker sandboxes with exec, process and terminal APIs, a sudo user, no SSH | SSH, root, systemd, Docker and /dev/kvm on every machine |
| Machine size | Not published; billed per browser hour | Browser size not published; sandboxes up to 32 vCPU and 256 GiB RAM, disk up to 32 GiB | 4 vCPU / 8 GB by default, up to 16 vCPU / 32 GB by plan |
| Session length | 15 minutes on Launch, 1 hour on Scale, up to 24 hours on Enterprise | Browsers up to 12 hours and sandboxes up to 24 hours, on every plan | No session cap on paid plans |
| Files after a session | Files API keeps session files; profiles keep cookies and logins (300 MB, deleted after 30 days unused) | Saved browser profiles; a stopped sandbox is gone unless files are on a volume or in a memory snapshot | The disk persists, with snapshots every minute and on stop; a stopped sandbox costs nothing |
| Snapshots and fork | No fork; a new session can start from a saved profile or session context | Memory snapshot of a running sandbox, then start many new sandboxes from it | Fork a running sandbox with its disk; no memory snapshots |
| Browser and computer use | Headful Chrome, a computer-use input API, an embeddable live view with human takeover, mobile mode and traces | Cloud Chrome over CDP with live view and recordings; hosted HyperAgent, Browser Use and computer-use models | 60fps desktop and Chromium, computer use and browser control, embeddable stream |
| Bot detection | Residential proxies, CAPTCHA solving and dedicated IPs on paid plans; bring your own proxy on every plan; stealth browser on Enterprise | Stealth mode, residential proxies and CAPTCHA solving on paid plans; ultra stealth and your own proxy on Enterprise | Datacenter IPs and Chromium; no CAPTCHA solving |
| Web data APIs | Browser tools for scrape, screenshot and PDF, billed per call | Fetch, Scrape and Search APIs per page or query, and hosted browser agents per step | None; your agent runs its own scraper on the VM |
| Reaching the machine | No inbound ports; a CDP websocket and a live view URL | HTTPS URL per exposed sandbox port, public or token-protected; no raw TCP or own IP | HTTPS URLs for up to 50 ports, token-protected; raw TCP; dedicated IPv6 or IPv4 |
| Secrets | Credentials API (beta) stores web logins and signs in without showing the agent the password | Keys are passed into sandbox processes as environment variables | AI credentials proxy: real API keys stay outside the sandbox, for model providers and any HTTPS API you register |
| Coding agents | A CLI, skills and an MCP server give your coding agent a Steel browser; the agent runs elsewhere | Base images with the Claude Code, Codex or OpenClaw CLI installed, run through the exec API | Claude Code, Codex, OpenCode and more preinstalled, with a prompt, steer and events API |
| Self-hosting | Open-source steel-browser (Apache-2.0) in Docker, and Steel Outpost inside your own network | Managed cloud only | Managed cloud only |
| Regions | us-east only; proxy location sets the visible IP | us-central, us-east, us-west and europe-west; asia-south on Enterprise and Select plans | EU only: Germany, Finland, France |
Every feature Battleships tracks (43)
| Feature | Steel | Hyperbrowser | boat |
|---|---|---|---|
| Full VM (own kernel) | Yes | Yes | Yes |
| Docker inside | No | No | Yes |
| Nested virtualization | No | No | Yes |
| Root | No | Yes | Yes |
| Systemd | No | No | Yes |
| Desktop GUI | No | No | Yes |
| Browser | Yes | Yes | Yes |
| Computer-use API | Yes | Yes | Yes |
| SSH | No | No | Yes |
| Persistent disk | No | No | Yes |
| Pause/resume | No | No | Yes |
| Automatic snapshots | Yes | No | Yes |
| Fork/clone | No | Yes | Yes |
| Live fork (no pause) | No | Yes | Yes |
| Memory fork | No | Yes | No |
| Custom image (Docker or snapshot) | No | Yes | Yes |
| ≥24 h sessions | Yes | Yes | Yes |
| Idle auto-stop | Yes | Yes | No |
| Wake on request | No | No | No |
| HTTPS preview URLs | No | Yes | Yes |
| Raw TCP inbound | No | No | Yes |
| Public IPv4 | No | No | No |
| Egress allowlist | No | Yes | No |
| Secret proxy | No | No | Yes |
| Secret proxy for any API | No | No | Yes |
| Preinstalled agents | No | Yes | Yes |
| Agent harness API | No | No | Yes |
| MCP server | Yes | Yes | No |
| GPU | No | No | No |
| Windows | No | No | No |
| Arm64 | No | Not published | No |
| Self-hosting | Yes | No | No |
| BYOC | Yes | No | No |
| Open source | Yes | No | No |
| Scoped API keys | Not published | Not published | Yes |
| Spend limits | Not published | Not published | Yes |
| Usage per sandbox (API) | No | Yes | Yes |
| Audit logs | Not published | Not published | No |
| SSO | Yes | No | No |
| SOC 2 | No | No | No |
| HIPAA | Yes | Not published | No |
| EU data residency | No | Yes | Yes |
| Zero data retention | Not published | Not published | Yes |
Performance
There is no comparable measurement for either. Neither Steel nor Hyperbrowser was in StarSling's public CPU benchmark, and neither publishes how many vCPUs a browser session gets, so there is nothing to set against boat's 26.69 runs per second at 4 vCPU / 8 GB. On start times, Steel says a session starts in under a second, and Hyperbrowser says under 500 ms for a ready browser and under 50 ms for a sandbox.
Node.js build-and-test loop at 4 vCPU / 8 GB, higher is better. Measured with the open hpc-sandbox-benchmarks suite. Steel and Hyperbrowser 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 | Steel, a month | Hyperbrowser, 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 | $250 · Scale · with limits | $16 · Free · with limits | $20 · $20/mo |
| One agent that never stops 4 vCPU / 8 GB, 50 GB disk, running 24/7 all month | $250 · Scale · with limits | $73 · Free · 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 | $784 · Scale · with limits | $792 · Scale · 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 | $510 · Scale · with limits | $450 · Scale · 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 | $950 · Scale · with limits | $994 · Scale · with limits | $2,000 · $2000/mo · with limits |
Steel: Steel bills browser hours by the minute, rounded up: $0.10 an hour on the free Launch plan and $0.08 on Scale, which has a $250 monthly fee that includes $100 of usage. Proxy bandwidth, CAPTCHA solves and browser tools are billed on top.
Hyperbrowser: Hyperbrowser spends one credit balance at $0.001 a credit: browsers cost $0.10 a browser-hour plus $10 per GB of proxy data, and sandboxes cost $0.045 per vCPU-hour plus $0.015 per GiB-hour. The Startup ($30) and Scale ($100) fees turn fully into credits each month.
Hyperbrowser: You pick a monthly plan from $20, and all of it becomes usage credit spent per second on running sandboxes. A stopped sandbox costs nothing and keeps its disk.
Limits in those scenarios
The Steel and Hyperbrowser prices in the table are for one browser session with no published size and no disk that keeps installed state, so neither fits any scenario as a 4 vCPU / 8 GB machine. Steel needs the Scale plan at $250 a month for sessions over 15 minutes and still stops them after 1 hour, while Hyperbrowser browsers run up to 12 hours on any plan. Both cap Scale at 100 sessions at once, so 300 sandboxes for your users and 1,000 short jobs at once need Enterprise on either. On boat the 1,000-job burst is the one that does not fit: bursts above 90 starts a minute are queued, and it needs more starts per day than the top plan allows.
| Scenario | Steel | Hyperbrowser | 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 Steel when
you want a focused browser API you can also self-host as open source, with a credentials vault, human takeover, traces and the lower $0.08 browser-hour rate on Scale.
- Open source you can host yourself. Useful if browsers must run on your own machines: steel-browser is Apache-2.0 open source, and Steel Outpost runs inside your network.
- Ready-made login and takeover tools. Useful for agents that sign in to sites: a credentials vault signs in without showing the password, and a person can take over the live view.
- Lower browser rate on Scale. Useful for heavy browser use: Scale bills $0.08 a browser-hour, and you can bring your own proxy on every plan.
Use Hyperbrowser when
your agent needs the web plus some code: stealth browsing, fetch and search APIs, 12-hour browser sessions, short sandboxes from memory snapshots, or regions beyond us-east.
- Sandboxes for code next to the browser. Useful if your agent also runs code: Firecracker sandboxes up to 32 vCPU, with memory snapshots and default-deny outbound rules.
- Longer sessions on every plan. Useful for long browser runs: sessions last up to 12 hours on any plan, where Steel stops at 1 hour on Scale.
- Web data APIs and more regions. Useful for teams outside us-east: fetch, scrape and search APIs per page, with regions in the US and Europe.
Where boat fits
Steel and Hyperbrowser are browser products, and both beat boat when the job is stealth scraping behind proxies and CAPTCHAs. If what you are really running is agents that browse and then install, build, run Docker or keep working for days, the browser is only one part, and boat gives each agent a full VM with Chromium, a desktop and coding agents on one machine.
A full VM with Docker and /dev/kvm
A Steel session has no shell, and Hyperbrowser's sandboxes do not document Docker or /dev/kvm inside. An agent that runs docker compose or an emulator needs a second machine next to either, which you wire up and pay for.
No session cap and a disk that stays
Steel stops a session after 1 hour on Scale and Hyperbrowser stops a sandbox after 24 hours, deleting its disk unless you used a volume or snapshot. On boat the disk is saved every minute and a stopped sandbox costs nothing.
Coding agents driven from outside, keys kept out
With Steel you host the agent yourself, and with Hyperbrowser you run its CLI through exec with the model key in an environment variable. boat ships the agents with a prompt, steer and events API, and its credentials proxy keeps real keys outside.
Use boat when your agent browses and also works: installs, builds, Docker and dev servers on a full VM with no session cap, a disk that stays, a 60fps desktop and coding agents built in. Compare it one to one: boat vs Steel and boat vs Hyperbrowser.
Questions people ask
Is Steel cheaper than Hyperbrowser?
It depends on the plan. Steel's Scale plan bills $0.08 a browser-hour against Hyperbrowser's $0.10, but costs $250 a month against Hyperbrowser's smaller fees. In the table, one coding agent on workdays comes to $250 on Steel against $16 on Hyperbrowser, and 50 agents in parallel to $784 against $792, with neither fitting.
Which is faster, Steel or Hyperbrowser?
Neither has a public CPU benchmark, and neither publishes the size of a browser session. Steel says a session starts in under a second, and Hyperbrowser says a browser is ready in under 500 ms.
Can Steel or Hyperbrowser run Docker?
A Steel session is a cloud Chrome with no shell, so there is nowhere to run Docker. Hyperbrowser builds custom sandbox images from Docker on your own machine, but its docs do not describe running Docker inside a sandbox. Every boat sandbox runs Docker, systemd and /dev/kvm.
How long can a session run on Steel and Hyperbrowser?
Steel sessions last up to 15 minutes on Launch, 1 hour on Scale and 24 hours on Enterprise. Hyperbrowser browser sessions last up to 12 hours and sandboxes up to 24 hours, on every plan. boat has no session cap on paid plans.
Can I self-host Steel or Hyperbrowser?
Steel, yes: steel-browser is open source under Apache-2.0 and runs in Docker, and Steel Outpost runs inside your own network. Hyperbrowser is a managed cloud only, as is boat.
Is boat SOC 2 compliant?
SOC 2 is in progress. We're open about where we stand: live status at status.ascii.dev and our data retention policy at docs.boat.dev/data-retention.
Try boat
Install the CLI, then run boat new.
7-day free trial