Overview
Squoosh is an open-source image compressor from Google Chrome Labs at squoosh.app. The GitHub repo GoogleChromeLabs/squoosh was created on 7 March 2018, is Apache-2.0, and is not archived. The site loaded on 2026-09-06. The latest commit visible on the default branch is 19 August 2024 (deprecated terser plugin; a single-thread AVIF fix shortly before). It is alive. It is not a SaaS with weekly releases.
The seller-relevant fact: codecs run in the browser via WebAssembly; the picture is not sent to a compression host. The README still discloses Google Analytics for visits, before/after byte sizes, and PWA install metadata. Unpublished factory packs can be tried as WebP/AVIF without landing on TinyPNG’s servers. The trade-off is just as sharp: one image at a time, no account, no bulk ZIP, no per-image invoice, no Squoosh Pro.
It does not matte. remove.bg owns alpha, Canva owns layout, TinyPNG owns lazy batch and WordPress. Squoosh’s job is: take an already-fixed main, ride the quality slider until kilobytes and zipper teeth agree, download.
Core features and advantages
Local codecs and a split view
Drop one image. Pick MozJPEG, OxiPNG, WebP, AVIF, JPEG XL, or whatever the current build exposes. Original on one side, a draggable divider in the middle. That is the whole point versus TinyPNG’s one-click blob: you can see whether a 40 KB AVIF melted the zipper. Resize, chroma, effort — the knobs are there for people who will use them. It is a codec playground, not a studio, and not a colour-management course.
WebP and AVIF for Shopify and Amazon
If a Shopify theme serves WebP/AVIF, Squoosh is how you check that a transparent PNG still has a clean edge after conversion. Amazon mains remain JPEG/PNG-first; the value is comparing JPEG quality 70 vs 85 on the same 2000 px white plate and seeing which one still zooms. Do not cut a white background here (no matte). Do not expect seven marketplace sizes (no artboards). Phones can run it; large AVIF will cook a mid-range Android. Desktop Chrome/Edge is the stable path.
CLI status and why it is not a pipeline
2026 writeups treat @squoosh/cli as unmaintained and point Node users at Sharp. This entry does not sell a dead CLI as a product. The web app is one file at a time. Two hundred SKUs × six images is an afternoon in a browser tab — TinyPNG’s API or a local script. The PWA can be installed; offline still depends on whether WASM is cached. Do not assume the first inflight open can encode AVIF.
Supported platforms
- Web at https://squoosh.app (modern browsers, WebAssembly)
- Source at https://github.com/GoogleChromeLabs/squoosh
- No official Amazon or Shopify plugin, no cloud account
- Raster in; not a video, PDF, or SVG converter
Marketplace rules do not change: Amazon mains want white and enough pixels; Shopify cares about LCP. Squoosh only emits a file.
Who it's for
- Solo sellers / designers finishing a handful of hero mains and wanting to see fabric. Free, no login.
- DTC sites deciding JPEG vs AVIF: run real product photos once, then put Sharp (or similar) in the build. Do not ask ops to click 200 times.
- Privacy-sensitive work: unreleased packs, return photos with faces, no SaaS upload.
- Skip it for catalog compression (TinyPNG API), white mattes (remove.bg / Photoroom), A+ modules (Canva), or a vendor SLA.
Concrete use case: a white PNG from Photoroom or remove.bg, 2000 px, 3 MB. In Squoosh, export JPEG q=80 and AVIF; inspect zipper and logo. Amazon gets the JPEG; Shopify gets AVIF/WebP. Write the settings down. Bulk goes to a script, not to 200 drags.
Pricing and value
Free. No plans. No 500-call quota, no annual seat. You pay CPU time and one-file labour. Optimal for five mains today; the wrong tool for eighty new ASINs tonight. “Made by Google” is not an enterprise contract. Chrome Labs experiment; issues live on GitHub.
Community feedback
Developer blogs and 2026 compression roundups still treat Squoosh as a quality reference (some authors publish pretty SSIM tables — their test set, not an SLA). Reddit and web-perf threads praise local processing and complain about batch. Seller forums almost never name it; it is not an ASIN tool. Chinese posts are usually “Google free compressor” shorts and rarely note that commits slowed after 2024. The site works; nobody is adding ecommerce features. Third-party Chrome extensions that queue files for Squoosh are third-party, not Chrome Labs.
Competitive positioning
TinyPNG: upload, batch, API bill, fewer decisions. Squoosh: local, tunable, slow.
remove.bg: matting. Squoosh will not drop a background.
Canva: canvas. If the export is still heavy, come back to Squoosh or TinyPNG.
Desktop ImageOptim and local Sharp scripts sit in the same “don’t upload” camp; they are not catalog links here.
Alternatives
- TinyPNG: batch work or WordPress auto-compress.
- remove.bg: cut the matte first, then compress.
- Canva: listing layout; a compressor still has to finish the file.
Limitations
One image at a time. Default-branch commits stopped in August 2024; features freeze. Do not run the CLI in production. Large AVIF burns CPU. Analytics still ships size numbers. No white background, no bulk, no ticket queue. Codec availability is whatever the live page exposes; JPEG XL is not guaranteed in every browser build.
Verdict
Keep Squoosh bookmarked in 2026 for a few mains, a visual compare, and files that must not upload. Do not clear a catalog with it. Next step: open squoosh.app, drop a real white main, export JPEG and WebP, check Amazon zoom and a Shopify preview. If the job is a zip of SKUs, use TinyPNG’s API or local Sharp — not a hundred download clicks.
Data notes
Rechecked 2026-09-06 against squoosh.app (live), the GitHub repo (created 2018-03-07, not archived, last default-branch commit 2024-08-19), and the README (local processing plus Analytics). monthlyVisits, domainRating, and authorityScore are catalog placeholders and unverified. Categories were corrected from Product Research to Operations / Marketing. Independent benchmark SSIM scores are not treated as official metrics here.