Superwhisper
All comparisonsComparison
Superwhisper vs Raycast
Raycast just added a dictation feature to its launcher. It's a solid first version. Superwhisper is a dictation-first product, with offline models, custom AI modes, file transcription, and compliance built in. Here's what we think is a fair comparison.
Key differences
Side-by-side comparison
| Superwhisper | Raycast | |
|---|---|---|
| Platforms | macOS, Windows, iOS | macOS, iOS (keyboard) |
| Offline mode | Full offline support | Cloud only |
| Privacy | On-device option, SOC 2, HIPAA | Cloud, audio not retained, no SOC 2 or HIPAA listed |
| Languages | 100+ with auto-detect and translation | Auto-detect, no published list |
| AI modes | Super, Voice, Email, Message, custom | Email, Messaging, custom styles |
| Reads screen context | Super Mode reads on-screen text | Reads frontmost app name only |
| Custom vocabulary | Yes | Yes |
| File transcription | Audio and video, with speaker labels | No |
| Meeting recording | Zoom, Teams, Discord, Slack | No |
| Enterprise | SSO, team billing, HIPAA, SOC 2 | Team plans, no compliance certs published for dictation |
| Product focus | Dictation-first | Launcher with dictation built in |
| Free tier | Free forever + 15 min Pro trial | Free during beta, expected to be Pro after |
Key differences
Where they actually differ
Offline vs cloud
Raycast dictation processes audio in the cloud and uses hosted models like OpenAI's GPT-5.4 mini to clean up transcripts. Their docs explicitly say local LLMs are not part of the current beta. If you lose internet, dictation stops.
Superwhisper runs its speech models on-device. You can work on flights, in coffee shops with bad wifi, or in secure environments where audio can't leave the machine. There are cloud models too if you want them, but the offline path is real.
Compliance
Raycast says voice isn't used to train models and audio isn't retained on their servers. That's a reasonable privacy posture. They don't publish SOC 2 Type II or HIPAA compliance for the dictation product today.
Superwhisper is SOC 2 Type II certified and HIPAA compliant. If you're in healthcare, legal, finance, or any regulated workflow, that matters more than a marketing claim.
Platform coverage
Raycast for Windows exists, but the new dictation feature is macOS plus an iOS keyboard right now. If you split your day across a Mac and a Windows machine, the dictation experience isn't consistent.
Superwhisper runs on macOS, Windows 10/11, and iOS. Same hotkey, same modes, same vocabulary, one license.
Modes and screen context
Both apps let you define styles for different contexts. Raycast ships Email and Messaging styles plus custom styles you can target per app.
Superwhisper's Super Mode reads what's on your screen to shape the output, not just the app name. If you're replying to a long email thread, it can pick up the thread context. You can also write modes with full system prompts.
What Superwhisper does that Raycast doesn't
File transcription with speaker labels. Meeting recording for Zoom, Teams, Discord, and Slack. A Claude Code integration. Translation between languages. Enterprise SSO and centralized billing. These aren't on Raycast's dictation roadmap today.
Where Raycast does well
If you already live in Raycast, having dictation in the same hotkey-driven app is genuinely useful. The launcher, AI chat, snippets, and now dictation all share one shortcut surface. The free beta also means you can try it without paying.
Raycast is a launcher that added dictation. Superwhisper is built around dictation. Different scope, both valid depending on what you need.
What people say
What people say about Superwhisper
"Superwhisper delivers on the dream of an AI-native operating system. The best part: it's insanely fast and does just what you expect."
"There's a new kind of coding I call 'vibe coding'... I just talk to Composer with @superwhisperapp so I barely even touch the keyboard."
Support
Frequently asked questions
Does Raycast dictation work offline?
No. Raycast dictation runs in the cloud and uses models like OpenAI's GPT-5.4 mini for post-processing. Their docs note that local LLMs are not part of the current beta. Superwhisper has full offline mode with on-device models, so audio never leaves your machine.
Is Raycast dictation available on Windows?
Not yet. Raycast for Windows exists, but the new dictation feature shipped on macOS and iOS first. Superwhisper runs on macOS, Windows 10/11, and iOS, and one Pro license covers every device.
Will Raycast dictation stay free?
It's free during the beta. Raycast has been clear that GPT-5.4 mini access and global dictation are usually paid features included for free during the beta, which suggests Pro will be required after launch. Superwhisper has a free tier that doesn't expire.
Which is better for developers and coding?
Both work in any IDE because they run system-wide. The difference is the rest of the stack. Andrej Karpathy coined 'vibe coding' using Superwhisper with Cursor, and Pieter Levels uses the same setup. Superwhisper also has a dedicated Claude Code integration. Raycast leans on its launcher commands and AI sidebar more than dedicated dev integrations.
What about file transcription or meeting recording?
Raycast dictation doesn't transcribe existing audio files or record meetings. Superwhisper handles both. You can drop in audio or video files with speaker labels, and record Zoom, Teams, Discord, or Slack calls.
How does compliance compare?
Superwhisper is SOC 2 Type II certified and HIPAA compliant, a requirement if you are handling regulated data. Raycast does not publish HIPAA or SOC 2 compliance for the dictation feature today.
Try Superwhisper free
Free tier that doesn't expire. 30-day refund if Pro isn't for you.
Download free