Every launcher comparison you find online was written by someone with a stake in the outcome. This one is no different, so we will start by naming ours and then try to earn back the credibility.
Our bias, stated
We make Vehla. We want you to install it. That incentive shapes which features get their own column in a matrix and which get folded into a footnote, and it is the reason most vendor comparisons are useless.
Two things we have tried to do about it. First, the matrix below uses columns that describe surfaces rather than products, so a tool that does one surface well is not penalized for the columns it leaves blank. Second, there are three sections naming the cases where we would tell you to use something else, and they are not throwaway paragraphs. If you only read one section, read Where Raycast wins.
One hard limitation to get out of the way, because it disqualifies us for a lot of people: Vehla requires macOS 15.0 or later on Apple Silicon. If you are on an Intel Mac or an older macOS, stop reading and check the current system requirements for the other two.
What each one is
Raycast is a launcher that grew into a platform. The palette is excellent, but the real product is the extension store and the developer ecosystem around it, plus an account layer that syncs your setup and, on the paid tier, provides hosted AI models. It is the most mature of the three by a wide margin and the only one whose extension catalog you could describe as an ecosystem rather than a directory.
SuperCmd is an open-source launcher, MIT-licensed, with a palette, clipboard history, window management, notes and canvas, and support for bring-your-own-key cloud providers plus Ollama. It is the answer for people whose first question is whether they can read the source.
Vehla is a command center rather than a launcher, in the sense that it deliberately absorbs surfaces that are usually separate apps: window management via SnapWheel, a notch surface, a file shelf, a menu-bar cleaner, clipboard intelligence, dictation, notes, canvas and Kanban. It ships an on-device MLX runtime so text actions can run with no network, and it is sold once rather than rented.
The feature matrix
Read this as a map of scope, not a scoreboard. A dash frequently means a deliberate decision rather than a gap.
| SURFACE | VEHLA | RAYCAST | SUPERCMD |
|---|---|---|---|
| Command palette | ✓ | ✓ | ✓ |
| Extension ecosystem | ✓ young | ✓ largest | ◐ |
| On-device model | ✓ MLX | ◐ via Ollama | ◐ via Ollama |
| Hosted AI (no key needed) | — | ✓ | — |
| Clipboard history | ✓ | ✓ | ✓ |
| Window management | ✓ | ✓ | ✓ |
| Notch / file shelf | ✓ | — | — |
| Menu-bar cleanup | ✓ | — | — |
| Dictation | ✓ | ◐ | — |
| Notes / canvas / boards | ✓ | ◐ | ✓ |
| Cross-device sync | ◐ iCloud | ✓ account | ◐ |
| Source available | — | — | ✓ MIT |
✓ built in · ◐ partial, add-on, or via a third-party provider · — not a core feature
The AI question
This is the only row in the matrix that reflects a genuine philosophical split, and it is worth more than a checkmark.
Raycast's paid tier gives you hosted models. You pay a subscription, you get access to frontier models, and there is no key management, no local download, and no capacity planning. For most people this is the least friction available, and pretending otherwise would be dishonest.
SuperCmd and Vehla both take the position that the model is your business, not the app's. You supply a cloud key, or you point the app at a local server, and the app never sits in the middle. The tradeoff is setup: you will spend fifteen minutes on a key or a download before anything works.
Where Vehla goes further is that it ships its own on-device runtime rather than only speaking to Ollama. Text actions can run against a bundled Gemma 4 model with the network off, and if the local model is unavailable Vehla shows an error instead of quietly falling back to a cloud provider. That last detail matters more than the first: silent fallback is the failure mode that makes "local mode" meaningless.
A 2B or 4B on-device model is not a frontier model. For rewriting, summarizing, extracting to-dos, and translation it is enough. For hard reasoning, long code, or anything requiring vision, it is not, and we route those to a cloud provider you choose. We wrote about exactly where the line falls in Local AI on your Mac.
Three-year cost
Prices are public US list prices as of 18 July 2026, for one person on one or two Macs, excluding whatever you spend on cloud API usage. Verify current numbers on each site before you commit.
| PRODUCT | MODEL | 3 YEARS |
|---|---|---|
| Vehla Free | Free forever | $0 |
| Vehla Pro | One-time, 2 Macs | $29.99 |
| Raycast free | Free core | $0 |
| Raycast Pro | $8/mo billed annually | ≈$288 |
| SuperCmd | MIT open source | $0 |
The honest reading of that table is not "Vehla is cheaper than Raycast." It is that Raycast Pro's $288 buys you hosted frontier models, and if you would otherwise pay for a chat subscription anyway, the comparison is closer to a wash. What a one-time license actually buys is the absence of a renewal decision.
Where Raycast wins
Pick Raycast if any of these describe you.
You want an extension for a specific SaaS tool. Raycast's catalog is the largest by a wide margin and the odds that someone has already built the thing you need are much higher there. Our Store is real, signed, and sandboxed, and it is also a fraction of the size. This gap will not close soon.
You want AI without configuring anything. Hosted models on a subscription is a genuinely better onboarding experience than downloading a 2.5 GB model file or pasting an API key.
You work across devices, or on a team. Raycast's account layer syncs setup and supports shared team configuration. Vehla syncs some customizations through iCloud and has no team story at all.
You are not on Apple Silicon with macOS 15+. Then this is not a comparison, it is a single option.
Where SuperCmd wins
Pick SuperCmd if the source code matters to you. An MIT license means you can audit what the app does with your clipboard, fork it if the maintainers make a decision you dislike, and never depend on a company staying in business. Vehla is closed-source, and no amount of privacy copy on our site substitutes for that.
It is also the right pick if you want a launcher and nothing more. Vehla's scope is a feature to some people and bloat to others, and if you already own Rectangle, Maccy, and Bartender and like them, a launcher that stays a launcher is the cleaner fit.
Where Vehla wins
Local by default, and it fails loudly. A bundled MLX runtime, models from 2B to 12B, and no silent cloud fallback. This is the case where we think the difference is not marginal: if your employer would need to run a privacy review before you install an AI tool, local mode is the difference between installing it and not.
The surfaces around the palette. SnapWheel, Island, Shelf, BarKeep, Folder Dock, clipboard image analysis, Flow dictation, Notes, Canvas and Kanban are all in the app rather than in eight other apps. If you look at the stack we overlap with and recognize most of it as installed on your Mac, that is the pitch.
One purchase. $29.99 once, two Macs, updates for life, and a free tier that is not a trial.
How to choose
If you need a specific integration or hosted AI with zero setup, Raycast. If open source is a requirement, SuperCmd. If your Mac currently runs four or five single-purpose utilities and you would rather run one app with a model that can work offline, Vehla. All three have a free tier that costs nothing but disk space, and an afternoon with each will tell you more than this page will.
How we checked
Feature claims and public US list prices were checked on each product's own site on 18 July 2026. Vehla details come from our docs for version 8.5, build 406. Competitor products ship frequently and this page will drift; treat it as a snapshot with a date on it rather than a live reference, and check the source before you act on a number.
Found an error? Email kailaconsulting@outlook.com and we will correct the post and note the change.
Common questions
Is Vehla cheaper than Raycast?
Vehla Pro is $29.99 once for two Macs. Raycast Pro is $8 per month billed annually, roughly $288 over three years. Both have a $0 tier. The honest reading is not that Vehla is cheaper: Raycast's $288 buys hosted frontier models, so if you would otherwise pay for a chat subscription the comparison is closer to a wash. A one-time license buys the absence of a renewal decision.
Which launcher has the best AI?
Depends what you mean. Raycast's hosted models are the least friction available. Vehla and SuperCmd treat the model as your business — a cloud key or a local server. Vehla ships its own on-device MLX runtime, so text actions run with the network off, and it errors rather than silently falling back to cloud when the local model is unavailable.
When should I choose Raycast over Vehla?
When you need an extension for a specific SaaS tool, when you want AI with no configuration, when you work across devices or on a team, or when you are not on Apple Silicon with macOS 15 or later — in which case Vehla will not run at all.
When should I choose SuperCmd?
When the source code matters. MIT licensing means you can audit what the app does with your clipboard, fork it, and never depend on a company staying in business. Vehla is closed-source. SuperCmd is also right if you want a launcher and nothing more.
Does Vehla run on Intel Macs?
No. Vehla requires macOS 15.0 or later on Apple Silicon. On an Intel Mac or an older macOS, the comparison is between Raycast and SuperCmd.