The best BackstopJS alternatives, compared honestly
BackstopJS still does the thing it was built for well: declare scenarios in a JSON file, screenshot them across viewports, review the diffs in a report with a before/after scrubber. But its last release was 6.3.25 on 7 September 2024, and nothing has been committed since — so the question is no longer "is it good?" but "what happens when Chromium moves and nobody is home?"
The best BackstopJS alternative depends on which half of the tool you actually want to replace. In short:
- Hosted baselines and approvals, in the pipeline that already deploys the site → Buddy — page-level visual tests with sitemap discovery and CSS/XPath ignore rules, no per-snapshot meter.
- Stay DIY, keep baselines in the repo, just on something maintained → Playwright's built-in
toHaveScreenshot()— free, and its browser binaries are pinned to the runner. - Cheapest hosted review with pricing you can actually read → Argos — free to 5,000 screenshots, then $100/mo for 35,000.
- Get the bitmaps out of git without leaving DIY → reg-suit — diffing and baseline storage only, bring your own screenshots.
- Component-level review of a Storybook → Chromatic, which is the wrong shape for whole-page testing but the right answer for components.
Why teams look elsewhere
What pushes teams off BackstopJS
Nothing here is a bug report. These are the structural properties of a free, config-driven, in-repo tool that stopped receiving releases — most of them were acceptable trade-offs while the project was moving.
Frozen since September 2024
Last npm publish and last commit are both 7 September 2024 (v6.3.25). The repo has 7,163 stars and 576 open issues, and is not archived — so it looks alive in search results while nothing ships.
The Puppeteer pin is ageing
v6.3.25 depends on puppeteer ^22.1.0 — a February 2024 line the caret will never move past — plus playwright ^1.40.1. Deprecated-dependency reports like issue #1586 (October 2024) are open and unanswered.
Baselines live in your repo
backstop approve copies bitmaps into backstop_data/bitmaps_reference/ and you commit them. Every scenario × viewport × approval is a binary blob in history: the repo grows and nobody can review the diff in a pull request.
You pin the rendering environment by hand
Diffs are only trustworthy if fonts, GPU and Chromium match between the machine that approved the baseline and CI. That's why the project ships a Docker path at all — and why "works locally, red in CI" is the classic BackstopJS false positive.
No shared review or sign-off
The scrubber report is genuinely one of the better diff UIs in open source, but it's a local artifact. There's no per-pull-request status, no reviewer approval, and no history for anyone who didn't run the suite themselves.
Headless Chrome only
One engine. Safari, Firefox and Edge regressions are out of scope by design — which is fine until a WebKit-only layout bug reaches production and the suite stays green.
The shortlist
7 BackstopJS alternatives worth trying
Ranked for the specific reader who already has a backstop.json and tests whole pages. Two lanes run through this list: stay DIY on something maintained, or hand baselines and approvals to a service. Pick your lane first — it matters more than the ranking.
Visual tests as a step in the pipeline that already builds and deploys the site: discovery from explicit URLs, a sitemap, crawled links or a URL file; regex URL excludes; ignore elements by CSS or XPath; delays and wait-for-element; scoped cookies and headers for authenticated pages. Same shape as a backstop.json, with hosted baselines and approvals instead of committed bitmaps — and no per-snapshot meter. Not a Storybook component service. Visual Tests action →
Free, Apache-2.0, and the most active tool on this page (1.62.0, 24 July 2026). toHaveScreenshot() ships in the runner and browser binaries are versioned with it, so the rendering environment is pinned for you. Weakness: baselines still land in git, and there's no review UI or approval workflow at all.
Free to 5,000 screenshots, Pro from $100/mo including 35,000, overage $0.004 ($0.0015 for Storybook), billed only on successful builds. Captures come from your own CI, so any framework works, and review lives on the pull request. Weakness: no self-hosting, and SSO is a paid add-on ($50/mo GitHub, $200/mo SAML).
Built by the Storybook maintainers; component-level snapshots and UI review are the best in the category. Free tier is 5,000 billed snapshots, Chrome only; Starter $179/mo for 35,000 plus Safari/Firefox/Edge, overage $0.008; Pro $399/mo for 85,000. Weakness that matters here: it requires Storybook — useless if you test deployed pages.
BrowserStack's visual testing service: mature cross-browser rendering, broad SDK coverage, free tier of 5,000 screenshots a month with 30-day build history. Weakness: paid prices aren't published on Percy's or BrowserStack's own pricing pages, so budgeting means a sales conversation — and pixel diffs still get noisy on dynamic content.
MIT, and the most surgical fix on this list: it doesn't capture anything. You keep your existing screenshots — even BackstopJS's own — and reg-suit handles diffing, baseline storage in your S3 bucket, HTML reports and GitHub notifications. Latest release 0.14.5 (26 August 2025). Weakness: you still own the capture half, and the cadence is slow.
Its "Visual AI" diffing genuinely cuts the false positives that make pixel-diff suites get ignored, and the Ultrafast Grid renders across many browsers and devices. Weakness: nothing is priced publicly — Starter, Public Cloud and Dedicated Cloud are all quoted in "test units" with typically annual contracts. Heaviest jump from a free CLI tool.
Also considered: Happo (free 5,000/mo Chrome-only; Starter $149/mo for 50,000; Growth $399/mo; Pro $749/mo; free for open source) is solid but component-driven and priced above Argos for the same job. Lost Pixel is not on this list on purpose — the team announced on 22 April 2026 that they are joining Figma and sunsetting the product, and the repository was archived the same day. Loki is Storybook-only and itself frozen since August 2024, and cypress-image-snapshot hasn't published since January 2021 — "open source" on its own is not a durability argument.
Side by side
BackstopJS alternatives compared
BackstopJS itself is in the table as the baseline you're measuring against. The maintenance column is the one that changed since 2024 lists — check it before you trust any older recommendation, including for the open-source options.
| Platform | Cost | Baselines | Review & approvals | Cross-browser | Licence · status | Best for |
|---|---|---|---|---|---|---|
| Buddy | Part of the CI/CD platform — no snapshot meter | Hosted | ✓ in-pipeline, with approved baseline | ✓ | Proprietary · active | Testing a deployed site from the pipeline that shipped it |
| BackstopJS (incumbent) | Free | Committed to your repo | ✗ local HTML report only | ✗ headless Chrome | MIT · frozen (Sep 2024) | Zero-cost, zero-account page diffing you pin and forget |
| Playwright | Free | Committed to your repo | ✗ none | ✓ Chromium, Firefox, WebKit | Apache-2.0 · active (1.62.0, Jul 2026) | Teams already running Playwright who just want a differ |
| Argos | Free to 5,000; $100/mo incl. 35,000; $0.004 overage | Hosted | ✓ on the pull request | via your own tests | Proprietary cloud, OSS SDKs · active | The cheapest honest exit from DIY, with published prices |
| Chromatic | Free 5,000 (Chrome only); $179/mo incl. 35,000; $0.008 overage | Hosted | ✓ component-level UI review | paid tiers only | Proprietary · active | Storybook design systems — requires Storybook |
| Percy | Free 5,000/mo + 30-day history; paid prices not published | Hosted | ✓ | ✓ | Proprietary · active | Broad SDK coverage inside the BrowserStack cloud |
| reg-suit | Free — you pay for your own S3 bucket | Your S3, not git | report + GitHub comment | via your own tests | MIT · active, slow (0.14.5, Aug 2025) | Keeping DIY capture but getting bitmaps out of the repo |
| Applitools Eyes | Quote only, "test units", ~annual | Hosted | ✓ | ✓ Ultrafast Grid | Proprietary · active | Large suites where false positives already killed trust |
Pricing models and free tiers change often — check each vendor for current terms. Compiled July 2026 from each vendor's official pricing pages; maintenance status from each project's GitHub and npm registry data, read 28 July 2026.
Official pages: BackstopJS · Playwright screenshots · Argos pricing · Chromatic pricing · Percy plans · reg-suit · Applitools pricing · Happo pricing · Lost Pixel sunset notice
Why we rank it first
What makes Buddy the strongest all-round pick
Most hosted visual testing is built for component snapshots taken by an SDK inside your test suite. A BackstopJS user isn't doing that — they declare pages, viewports and ignored regions and diff a deployed site. Buddy's Visual Tests action is built around exactly that shape, and it runs as a step in the pipeline that just deployed the thing being tested.
Pages come from the sitemap
Discovery from explicit URLs, a sitemap, crawled links or a URL file — instead of hand-maintaining a scenario array as the site grows. Robots.txt is respected during discovery.
Ignore regions by CSS or XPath
The direct equivalent of BackstopJS selectors/hideSelectors, plus URL exclusion by regular expression, so carousels, timestamps and ad slots stop producing red builds.
Delays and wait-for-element
Waits for a specific element or a fixed delay before capturing — the onReady problem, handled in configuration rather than in a script you maintain.
Authenticated pages
Scoped cookies and custom HTTP headers per request, so staging behind basic auth or a session cookie is testable without a login script.
Same pipeline as build and deploy
The diff runs after the deployment step that produced the change, against the environment it produced. No second CI system, no artifact hand-off, no separate credentials.
No per-snapshot meter
Baselines and approvals are hosted, and adding viewports doesn't multiply a bill — the failure mode where teams quietly cut coverage to stay inside a snapshot quota doesn't apply.
Where it isn't the answer. Buddy tests deployed pages, not Storybook component snapshots — if you want component-level review, Chromatic is genuinely better and we've ranked it accordingly. And BackstopJS asks for nothing: npx backstopjs test runs on a laptop, offline, with no account. Buddy is a platform you adopt. If all you want is a differ, Playwright or reg-suit is the honest recommendation.
A fair call
When BackstopJS is still the right choice
Frozen is not the same as broken, and several of the reasons to leave flip into reasons to stay depending on who's running the suite.
BackstopJS is fine if…
- It costs nothing and asks for nothing — no account, no vendor, no meter, no network call. For a small site or an agency handoff that's a feature, not a compromise.
- You can pin it.
backstopjs@6.3.25plus a pinned container image behaves identically next year; a tool that stopped changing also stopped introducing surprises in your pipeline. - One reviewer looks at the diffs. The scrubber report is still one of the best diff UIs in open source, and nothing in the paid tiers obviously beats it for a solo reviewer.
- You want provenance. In-repo baselines mean the approved state is in git history, reviewable and revertable alongside the code that changed it.
Consider an alternative if…
- More than one person needs to approve visual changes, with a status on the pull request — that's Argos at the low end, Buddy if the diff should live in the deploy pipeline.
- The repository is getting heavy with bitmaps, or reviewers can't see what a "baseline update" commit actually changed — reg-suit fixes exactly this without abandoning DIY.
- You've been burned by local-versus-CI false positives: Playwright pins browser binaries to the test runner, which removes the class of problem rather than dockerising around it.
- You need Safari, Firefox or Edge coverage, or AI-assisted diffing to stop a noisy suite being ignored — Percy and Applitools respectively, and neither is cheap.
- The Puppeteer 22 pin blocks a Chromium capability you need, and you'd be maintaining a fork to get it.
Common questions
BackstopJS alternatives — common questions
Is BackstopJS still maintained in 2026?
Not actively. The last npm release is 6.3.25, published on 7 September 2024, and the GitHub repository has had no commits since that same day. There are 576 open issues, including unanswered reports about deprecated dependencies. The project is not archived and not deprecated — it still installs and still works — but nobody is currently shipping fixes, so treat it as frozen rather than dead. (Some 2026 listicles still describe it as "actively maintained"; that contradicts the repository's own history, so check the commit log yourself.)
What is the closest drop-in replacement for BackstopJS?
If you want to stay DIY with baselines in the repository, Playwright's built-in toHaveScreenshot() assertion is the closest match and the most actively maintained option — version 1.62.0 shipped on 24 July 2026, and its browser binaries are versioned with the test runner, which removes the local-versus-CI rendering mismatch that causes most BackstopJS false positives. If you want hosted baselines and a review workflow instead, Argos is the cheapest option with published pricing.
How do I stop committing baseline screenshots to git?
Two paths. Stay DIY and add reg-suit, which keeps baselines in your own S3 bucket and only handles diffing, baseline management and reporting — it does not capture screenshots, so you keep your existing capture step. Or move to a hosted service (Argos, Percy, Chromatic, Happo, Applitools), where baselines and approvals live in the vendor's dashboard and nothing binary lands in your repository.
Isn't Lost Pixel the open-source alternative everyone recommends?
Not any more. On 22 April 2026 the Lost Pixel team announced they are joining Figma and sunsetting the product; the GitHub repository was archived the same day. Lists written in 2024 and 2025 still point BackstopJS users at it, so check the repository before you migrate. The announcement gives no sunset date, no migration path and no statement about existing subscriptions or data.
Do I need Storybook to use these tools?
Only for Chromatic, which is built around Storybook and cannot run without it. BackstopJS, Playwright visual comparisons, Argos, Percy, reg-suit and Applitools all work on whole pages or arbitrary screenshots. If you are testing a deployed site rather than isolated components, a Storybook-first tool adds a build step you do not need.
What is the cheapest hosted option with published pricing?
Argos: a free Hobby tier up to 5,000 screenshots, then Pro from $100/mo including 35,000 screenshots with $0.004 overage ($0.0015 for Storybook screenshots), billed only on successful builds. Chromatic's first paid tier is $179/mo for 35,000 snapshots and Happo's is $149/mo for 50,000. Percy publishes a free tier of 5,000 screenshots per month but does not publish its paid prices. Applitools quotes everything through sales.
Can I keep BackstopJS instead of migrating?
Yes, and for some teams that is the right call. Pin backstopjs@6.3.25 and pin the container image it runs in, and the suite will behave the same next year — a tool that stopped changing also stopped introducing surprises. The realistic upgrade path is incremental: keep the capture step, add reg-suit to move baselines out of git, and only replace the runner when the Puppeteer 22 pin actually blocks a Chromium feature you need.