Orion browser on a Mac, and what it trades away

Most write-ups about Orion stop at the headline: a WebKit browser that runs Chrome and Firefox extensions, built by Kagi, with no telemetry. All of that is accurate and none of it answers the question a working Mac user is actually asking, which is what breaks after the switch. Every browser choice is a trade. The useful version of this comparison names both halves.

The short version of the trade

Orion is built on WebKit, the same engine Safari uses, and its own documentation states the difference from Safari in one line.

Unlike Safari, Orion has native support for both Chrome and Firefox extensions. Source: help.kagi.com

That single design decision produces everything worth knowing about the browser. On the gain side: Safari-class resource behaviour on Apple silicon, ad and tracker blocking on by default, zero telemetry, and installable extensions from three different stores. On the cost side: a rendering engine that a meaningful share of business web applications were never tested against, and an extension catalogue that is curated rather than open.

The download page makes the reach obvious. Builds are published for macOS 26.0, 15.0, 14.0, 13.0, 12.0, 11.0, 10.15 and 10.14, which is an unusually long tail for a browser still in active development. A 2017 MacBook Pro stuck on an older system can run a current browser build, and that alone decides the question for some readers.

Pricing sits outside the feature set entirely. The browser itself is free. Orion+ is sold at 5 dollars a month, with a lifetime licence at 150 dollars, and the listed benefits are custom app icons, a decorative window border, a community tag and early access to beta builds. No part of the core browsing experience sits behind that wall, which is rare enough to be worth stating plainly.

What the WebKit base actually buys

Two things, and they are not the same thing.

The first is resource behaviour. Kagi publishes a specific claim about it rather than a vague one.

On default settings, Orion will consume 2x-3x less memory than mainstream choices after closing tabs. Source: help.kagi.com

Note the qualifier at the end. The claim is about memory reclaimed after closing tabs, which is a different measurement from steady-state footprint with twenty tabs open. It is a reasonable claim to make and an easy one to misread. Anyone evaluating on this basis should open Activity Monitor, sort by memory, and watch the same workload in both browsers for a full working day rather than trusting either side's number.

The second is privacy posture without configuration. Ad and tracker blocking is native, not an extension, and the product describes itself as zero telemetry. For a machine that handles client accounts, the practical value is that the default state is the safe state. Nothing has to be installed, and no setting has to be remembered on a fresh install.

There is a third effect that gets less attention. Because blocking happens inside the browser rather than through an extension, the usual first-run ritual of installing a content blocker and waiting for it to sync filter lists disappears. On a slow connection or a locked-down machine, that matters more than it sounds.

What the WebKit base costs

Compatibility is the real bill, and it arrives unevenly.

Consumer sites are fine. Anything that works in Safari works here, because it is the same engine. The problem is the category of software that only ever gets tested in Chrome: payroll portals, expense systems, bank corporate interfaces, older intranet applications, industry tools written against a specific Chromium behaviour. These do not fail loudly. They fail in one screen, usually a file upload, a print preview, or a signature pad, and usually two weeks after the switch when the previous browser has already been removed.

The mitigation is a list rather than a hope. Before committing, open every business-critical web application once, and specifically exercise the steps that involve uploading, downloading, printing or signing. Half an hour of that is worth more than any review.

Extensions carry a second cost that is easy to misread as a limitation and easy to misread as unlimited. Orion supports extensions from the Safari, Chrome and Firefox ecosystems, and publishes a gallery of curated, verified ones. The catalogue names entries including 1Password, uBlock Origin, Dark Reader, Consent-O-Matic, Tampermonkey, Mullvad VPN, iCloud Passwords and Microsoft Single Sign-on. Outside that curated set, an arbitrary Chrome extension may work, partially work, or not install. The check is three items long: the password manager in use, whatever the employer mandates, and the one extension that would end the trial if it were missing.

Platform reach is the last cost. macOS is the flagship, iOS and iPadOS are shipping, Linux is in beta, and Windows is described as an alpha in active development. A person whose desktop is a Mac and whose phone is an iPhone loses nothing. A person who also sits at a Windows machine two days a week does not get the same browser there yet.

Containers and profiles do two different jobs

This is the part of Orion that maps onto real work, and the two mechanisms are frequently confused.

Containers separate cookies and login state inside one browser window set. The documentation states the purpose directly.

Containers enable users to separate specific activities into color-coded containers. The main benefit of containers is that they are isolated from one another, allowing the user to use sites (e.g., email) while logged into different accounts. Source: help.kagi.com

Profiles go further down. Settings, passwords, extensions, bookmarks, history and site data split at the application level, and each profile appears in the Dock as a separate application with its own icon. That last detail is the one that changes daily behaviour, because switching stops being a menu action and becomes a Dock click.

The rule of thumb that falls out of this: containers for two or three accounts on the same service, profiles when the separation needs to include different extensions and different passwords. The failure case is using containers for a separation that really needed profiles, then wondering why a work extension is leaking into personal browsing. The same distinction, expressed as permanently separated work areas rather than as browser settings, is what a dedicated Workspaces model is built around, and comparing the two is the fastest way to tell which level of separation a given day actually needs.

What Orion does not try to solve

Two problems sit outside its scope by design, and neither is a defect.

Notifications stay with the individual sites and the macOS Notification Center. There is no single place that says which of eleven web applications made a sound, and no browser-level schedule that silences all of them between 18:00 and 09:00. For a reader whose main complaint is notification scatter rather than browsing, this is the wrong layer to fix it at.

Window sprawl is the second. Orion can install any site as a standalone app, with its own Dock icon, its own start URL, an option to float on top, and a choice of which profile it runs under. That is genuinely useful, and it produces one window per app. Install ten and there are ten windows, at which point the switching problem has moved from tabs to the macOS window manager rather than disappearing. A design that keeps every app permanently mounted in one window and switches between them with a keystroke is a different shape of answer, and the Features list is the concrete version of what that shape includes.

Neither gap is hidden. The point of naming them is that a reader who arrived because of notification noise will not get relief from an engine swap, and should stop evaluating engines.

A one week test that produces a decision

Downloading a browser and using it casually produces an opinion, not a decision. A test with a defined shape produces the second one, and a week is enough.

Day one is import and inventory. Bring bookmarks and passwords across, then write down every web application the job depends on. Most people arrive at somewhere between eight and fifteen, and the list is shorter than expected once it is written rather than remembered.

Days two and three are the compatibility pass. Open each item on that list and exercise the awkward steps specifically: upload a file, download a report, print a preview, sign something, paste a large block of formatted text. These are the operations where an engine difference shows up, and they are the ones nobody tests before switching.

Day four is the extension pass. Install only the non-negotiable extensions and use them normally. A password manager that fills correctly on eight sites and silently fails on the ninth is the common failure, and it takes a day of ordinary work to find.

Days five to seven are the account pass. Set up containers for the services with two accounts, profiles for the separations that need different extensions and different passwords, then work normally and count how often the wrong account appears. That count is the answer. If it drops to near zero the setup fits, and if it stays where it was the separation was drawn at the wrong level.

Anything still unresolved at the end of the week is usually a question the vendor has already answered in writing, and the FAQ page is a faster place to look than a forum thread.

Where it sits against the alternatives

Published free ceilings and prices, current as of this writing.

Product	Engine	What the free tier allows	Paid
Orion	WebKit	The entire browser	5 dollars a month, 150 dollars lifetime
Wavebox	Chromium	2 groups, 2 apps in each, 1 extension	8.33 dollars a month billed annually, Teams at 12.50 per user
Shift	Chromium	5 spaces, 10 apps per space	199.99 dollars a year, 19.99 a month
Rambox	Chromium	Unlimited apps, 2 instances per app	7 dollars a month, 70 a year
Ferdium	Chromium	No limits, Apache-2.0 open source	None

The middle column is the one that decides things. Rambox limits sessions rather than apps, which fails immediately for someone with three accounts on the same service. Wavebox limits groups and apps per group. Shift limits both, more generously. Orion limits nothing, because it is not trying to be an app aggregator at all, and its membership is a support mechanism rather than a feature gate. Reading the Pricing page of any candidate in this category with the free ceiling next to the monthly figure, rather than the monthly figure alone, is what turns a price list into a decision.

Also worth noting: this category churns. Sidekick, which appeared in comparison lists for years, was acquired by Perplexity and its site now redirects elsewhere. Checking the dated release notes and current ownership of any candidate takes ten minutes and prevents a second migration a year later.

What to change first

Pick the one sentence that describes the actual problem. If it is "Safari is fine but the extension is missing", Orion answers that directly and the trial can start today. If it is "the same service needs three accounts open at once", test containers and profiles for a week before deciding. If it is "the noise and the hunting", the engine is not the variable, and a browser built to keep each app in its own permanent space is the thing to compare against, starting with SpaceDeck.