Steel.dev

Steel.dev — Open-source browser API for AI agents, hosted or self-run. Steel publishes its entire browser service under Apache-2.0, which makes it the only credible middle ground between a hosted agent host and rolling your own. The hosted version adds a session viewer, proxy egress and managed scaling; the open-source image gives you the same session API on your own hardware. Its cookbook of integrations with browser-use, CUA loops and LangChain is where most people meet it. Founded 2024. Seed. Open source under Apache-2.0.

We track it in 1 comparison — Browser APIs — so every claim below is a cell in a table you can open and check rather than an impression. Across those rosters it sits against 12 other tools, and what follows is where it visibly separates from them.

Where it wins.

• Open-source core — Yes, which only 3 of 13 tools here manage.

Each of those is ranked against the whole roster on its category page, not against a hand-picked subset, so a first place here means first of everything we list.

Where it gives ground.

• GitHub stars — 5,000, the weakest of the 5 here. Self-hosted Playwright records 73,000.
• Funding stage — seed, 8th of 13. Cloudflare Browser Rendering records public.

None of these disqualify it on their own. They are the fields to check against your own requirements before you commit, because they are the ones where a competitor genuinely does better.

Provenance. 21 of 26 tracked fields carry a value for Steel.dev, and 7 of those cite a document you can open. Last verified 2026-07-01. Every figure keeps its own provenance — measured by us, claimed by the vendor, inferred, or community-reported — and we would rather print a dash than a guess.

Its nearest neighbour in our data is Airtop. Steel.dev is ahead on Open-source core (Yes against No) and CDP endpoint (Yes against Partial). Airtop takes CAPTCHA solving (Yes against Partial) and Agent API (Yes against Partial). That pattern repeats across the rest of the roster — see Steel.dev alternatives for the other rivals, each compared the same way.

Founded

2024

Funding

Seed

Licence

Apache-2.0 (open source)

Fields we track

21 of 26

Last verified

2026-07-01

The only option that removes vendor risk entirely — if the company disappears, the Apache-2.0 image keeps running. The hosted product is thinner than Browserbase on polish and the anti-bot stack is less battle-tested, but for agent work on cooperative sites the trade is a good one.

Open-source core

●Yes1st of 13

Vendor-claimedverified 2026-07-01source

GitHub stars

5,0005th of 5

Inferred

Stealth mode

●Yes1st of 13

Vendor-claimedverified 2026-07-01source

CDP endpoint

●Yes1st of 13

Vendor-claimedverified 2026-07-01source

Session persistence

●Yes1st of 13

Vendor-claimedverified 2026-07-01source

Playwright

●Yes1st of 13

Vendor-claimedverified 2026-07-01source

Live view

●Yes1st of 13

Vendor-claimedverified 2026-07-01source

Puppeteer

●Yes1st of 13

Vendor-claimedverified 2026-07-01source

Steel.dev is the right call in these situations, each one drawn from a field we actually record:

• Teams with a hard self-host requirement.
• Prototyping agents you may later run on your own hardware.
• Avoiding lock-in without writing your own session manager.
• Open-source core is your binding constraint. Steel.dev records Yes, which only 3 of the 13 tools in the Browser APIs roster do. We define that field as a self-hostable core under a public licence, so you can leave.

Reach for something else when any of the following is a requirement rather than a nice-to-have:

• GitHub stars. Steel.dev records 5,000, the weakest of the 5 tools in the Browser APIs roster. Self-hosted Playwright records 73,000 on the same field. If that is a hard requirement rather than a preference, start elsewhere.
• Funding stage. Steel.dev records seed, 8th of 13 in the Browser APIs roster. Cloudflare Browser Rendering records public on the same field. If that is a hard requirement rather than a preference, start elsewhere.

We publish this block because a comparison that only lists what a tool is good at is marketing. Every figure above sits on the same page as its source, and the field definitions are on the category tables if you want to check how we measured them.

What we don't know. 5 of the 26 fields we track for Steel.dev are still blank: Anti-bot pass rate, Cold start, Max concurrency, Chrome extensions and Per browser-hour. Those render as dashes rather than as zeroes or assumptions, because an empty cell and a bad cell are not the same thing and only one of them is honest. If you know any of these figures and can point at a document, tell us.