The best macOS launcher in 2026: a tested, no-AI-hype roundup

A modern macOS launcher should do four things well: open apps and files in under a quarter second, run a few high-impact actions without leaving the keyboard, stay out of the way when you do not need it, and never quietly ship your text to a server you do not control. That last point matters more than it used to. Most launchers in 2026 are either AI-focused (Raycast Pro), maintenance-mode (Alfred), or one-person-FOSS (Ueli). Picking is harder than it should be.

I have used every macOS launcher worth using in the last decade. This is the 2026 roundup, ordered by how much I would actually recommend each one for the persona it fits. No paid placements, no affiliate links, no "10 best" filler — only six options because that is how many are still maintained and worth installing.

What "best" means here

I scored each launcher on five axes that map to the work most developers, writers, and ops folks do all day:

1. Cold-search latency — first keystroke to first result. Under 30 ms feels instant, 30–80 ms feels usable, over 80 ms feels broken.
2. Power-user actions — kill-by-port, window management, clipboard history, snippet expansion, file actions.
3. Privacy posture — what leaves your machine, what the privacy policy actually permits, who funds the company.
4. Pricing model — one-time vs subscription, upgrade math over three years, free-tier limits.
5. Maintenance health — last release, active bug fixing, named maintainers.

Speed and privacy are non-negotiable for me. The other three I weight differently per persona.

The roundup

1. CmdSpace — best overall for indie devs and privacy-conscious users

Pitch: a local-only macOS launcher that runs every query on-device, ships with a developer command kit, and charges $29 once.

If you spent the last year frustrated that Spotlight regressed again (eclectic light covers the long arc), or that Raycast started routing your queries through a cloud LLM by default, CmdSpace is the obvious next stop. It binds to Cmd+Space, returns first results in well under 25 ms on an M1 Air, ships with a small but well-chosen command kit (kill-by-port, clipboard history, snippet expansion, window management, calculator with currency conversion), and does it all without sending a single packet to a CmdSpace server.

Trade-offs: no AI assistant, no team sync, no extension marketplace yet. The roadmap is conservative. The team is small enough that you actually get replies. If you want a launcher that grows but stays local, this is the most defensible pick in 2026.

Pricing: $29 one-time, 60-day trial. Major versions are paid upgrades at roughly 50% of list price, but only when there is a major version to ship.

Best for: P1 indie developers, P2 privacy-conscious power users, P3 Spotlight refugees who want one more app, not a workflow rewrite.

2. Raycast — best for teams already on a subscription stack

Pitch: the most feature-complete launcher in 2026, with deep team features and a polished extension ecosystem.

Raycast is what most launcher reviewers default to recommending, and there is a reason: it is genuinely good. Window management, snippets, clipboard, calendar peeking, hundreds of community extensions, a healthy keyboard model. The free tier is generous if you do not need AI or team sync.

The complication is that Raycast Pro is now $8/month or $96/year (raycast.com/pricing), and an increasing share of the marketing real estate is dedicated to Raycast AI which routes your prompts through their backend to OpenAI/Anthropic. The privacy page is reasonable about this — they tell you exactly what gets sent. But the model is "trust us with your context" rather than "your data never leaves your Mac." For a lot of teams that trade is fine. For a privacy-first solo user, it is the wrong direction.

Trade-offs: subscription pricing compounds; AI features are the strategic priority going forward; some power features are pro-only.

Pricing: Free for individuals (most features), $8/mo Pro for AI + sync, team plans on top.

Best for: teams that already pay for SaaS launchers and want a single tool everyone uses.

3. Alfred — best for muscle memory you cannot give up

Pitch: the original Mac launcher, still maintained, v6 era as of 2026.

If you have a 100-item Alfred workflow library and a Powerpack license from 2014, you do not need a roundup post to tell you whether Alfred is good. It is. It is fast (under 40 ms typical), the workflow system is unmatched for power users who can write a bash script, and the license is a sane one-time fee with paid major-version upgrades.

The honest assessment in 2026: Alfred's release cadence has slowed considerably. Big features land yearly, not monthly. The marketing site looks like it was last refreshed in 2019. The team is small and quiet. None of that makes Alfred bad — plenty of software is best when it stops thrashing — but it does mean Alfred is no longer the launcher pushing the frontier. If you switched from Alfred to Raycast in 2022, you will not feel a strong pull back.

Trade-offs: UI shows its age; ecosystem is graying; team is opaque about roadmap. For the full read on where Alfred actually stands today, see Alfred v6 status in 2026.

Pricing: Free for basics, £34 single license for Powerpack (one-time), £59 mega supporter, paid v→v upgrades.

Best for: P1 long-time Alfred users with workflow libraries they cannot replace.

4. Ueli — best fully open-source option

Pitch: an Electron-based open-source launcher with a credible community and zero pricing pressure.

Ueli is the OSS choice for users who refuse to install closed-source software in their core hotkey chain. It is cross-platform (macOS and Windows), respectfully feature-rich (search, calculator, file actions, plugin system), and shipped under an open license. The maintainer is responsive on GitHub.

The trade-off is that Ueli runs on Electron, which means an idle memory footprint around 150–200 MB and cold-launch latency a tier slower than native launchers. For a lot of users that does not matter. For the M1 Air user who notices every animation, it does.

Trade-offs: Electron overhead; smaller plugin ecosystem; less polish in animations and edge cases.

Pricing: free, open source, donations welcomed.

Best for: P2 privacy users who weigh open source above raw speed.

5. LaunchBar — best for keyboard purists who liked it 15 years ago

Pitch: Objective Development's LaunchBar has been quietly shipping since 1995, and the 2026 version still has the cleanest keyboard model in the category.

LaunchBar's "abbreviation search" — typing "go" gets you Google, typing "sl" gets you Slack — is the fastest mental model I have used for hitting apps without taking my hand off home row. The clipboard history is built in. The Instant Send model (Cmd+Tab to a target app while results are highlighted) is unmatched.

The honest take in 2026: LaunchBar feels old. The UI is dense and unapologetic about it. The community is small but loyal. Updates land but rarely. If you can tolerate the aesthetic, LaunchBar will outlast most of the apps you install this year.

Trade-offs: dated UI; small community; pricing model is one-time but expensive for what most users will use.

Pricing: €29 single user, €19 upgrade.

Best for: P1 keyboard purists who never left and never will.

6. Quicksilver — best for nostalgia, hardest to recommend

Pitch: Quicksilver is the OG, still maintained by a community of true believers.

Quicksilver introduced the noun-verb-object grammar that every modern launcher copies. It is still free, still open source, and still works on modern macOS — but the codebase shows its age in dozens of small ways, and the user community is now thin enough that asking for help on Reddit usually returns silence.

I cannot recommend Quicksilver as a primary launcher in 2026 for anyone who does not already use it. If you do use it, you do not need this post.

Trade-offs: dated; sparse community; rough edges on recent macOS releases.

Pricing: free, open source.

Best for: users with a decade-plus of muscle memory and no desire to relearn.

Side-by-side at a glance

Numbers are anecdotal from my M1 Air, not lab-grade benchmarks — but they reflect what you will feel.

How to decide in 60 seconds

• You are a privacy-first solo dev or writer: CmdSpace. Skip the rest of this section.
• You manage a team that already pays for tools: Raycast Pro. Their team features are worth the subscription.
• You have a Powerpack and 50 workflows: stay on Alfred. Migration cost is real.
• You refuse closed source: Ueli.
• You want the cleanest keyboard model and like dense UIs: LaunchBar.
• You have used Quicksilver for 15 years: you are not reading this.

What I did not cover

Two categories deliberately left out:

1. Spotlight itself. Apple's own launcher has regressed in measurable ways in Sequoia and Tahoe — see our migration guide for the specifics on indexing storms and the Tahoe 26 changes. The summary: Spotlight is fine as a system tool, no longer fine as your primary launcher.
2. AI-focused standalone launchers like various Claude/Copilot wrappers. They are interesting, they are not yet stable, and they all route through someone else's GPU. Wait six months.

The pick

If you want one answer: CmdSpace for solo users who care about speed and privacy, Raycast for teams who want everything in one app. Everything else is a special case.

If you are migrating off Spotlight specifically, the step-by-step migration guide is the next read. If you want to know how CmdSpace compares directly to Raycast or Alfred, the CmdSpace vs Raycast and CmdSpace vs Alfred deep-dives are the next two reads after that.

Sources

• Raycast pricing and Pro page — raycast.com/pricing, raycast.com/pro
• Raycast AI and privacy posture — raycast.com/core-features/ai, raycast.com/privacy
• Alfred status and pricing — alfredapp.com
• Ueli open-source launcher — ueli.app
• LaunchBar product page — obdev.at/products/launchbar
• Quicksilver project — qsapp.com