Skip to content
Petal

ComparisonsMac5 min read

Petal vs Superwhisper for Mac

Petal vs Superwhisper: both run local models on Mac. Compare own-key cloud models, modes vs the per-app Router, clipboard handling, setup, and license.

Published by Optimal Apps, the maker of Petal.

On this page

Petal vs Superwhisper is a close match, because both apps run speech models locally on a Mac. The main differences are price and structure. Superwhisper's free plan has unlimited local Whisper models, and its Pro plan adds cloud models and your own API keys. Petal for Mac is free and open source, and your own OpenAI, Anthropic, OpenRouter, Ollama, or LM Studio key works at no cost. Superwhisper organizes behavior into modes. Petal uses a per-app Router. Superwhisper also runs on Windows, iOS, and Android.

Disclosure: Optimal Apps publishes Petal and this site. The Superwhisper facts here come from superwhisper.com and its documentation at the time of writing.

Petal vs Superwhisper: quick comparison table

Superwhisper Petal for Mac
Price Free; Pro $8.49/month, $84.99/year, or $249.99 lifetime Free
License Closed source MIT, open source
Local models Unlimited Whisper models on Free Parakeet, Whisper, Qwen3 ASR, Voxtral, Nemotron, Apple Speech
Own API keys Pro Free
Cloud models Pro Your own key, free
Audio and video file transcription Pro Free
Behavior per context Modes (Message, Email, Formal, Casual, Legal, Chat, Custom) Router: route per app or website
Clipboard "Clipboard content" setting, "Simulate keypresses" Restore after 700 ms, or turn off Restore Clipboard
Platforms macOS, Windows, iOS, Android macOS; iPhone and Windows are separate Petal apps
Intel Macs Works best with cloud models Not supported; Apple silicon only
Refund 14 days Not needed

Local models in both apps

Both apps use local models, so both can work with no network after the download.

Superwhisper runs Whisper models locally on the free plan. Its FAQ says Intel Macs "work best with Cloud models", because local models need Apple silicon to be fast.

Petal runs models through FluidAudio (Core ML), WhisperKit, and MLX. In Settings > Models, you can download:

Model Size Notes
Parakeet 110M about 455 MB Recommended, fast English
Parakeet 0.6B V3 about 485 MB 25 European languages
Whisper Large V3 Turbo about 1.1 GB 99 languages
Qwen3 ASR 1.7B about 2.5 GB 30 languages, high accuracy
Parakeet Unified 0.6B, Nemotron Unified 0.6B about 625 MB, 1.2 GB Live text as you speak
Voxtral Realtime 4B about 3.2 GB Streaming, 13 languages

Petal keeps the default models under 8 GB of RAM. Petal builds only for Apple silicon and needs macOS 26.2 or later. If you have an Intel Mac, Superwhisper with a cloud model is the better choice.

Own-key cloud models: paid vs free

This is the clearest difference in price.

In Superwhisper, the pricing page lists "use your own AI API keys" and cloud models under Pro. You can try Pro features for 3,000 words.

In Petal, open Settings > Intelligence and choose Cloud Model. Then:

  1. Pick a provider: OpenAI, Anthropic, OpenRouter, or Custom Server (for Ollama, LM Studio, or any OpenAI-compatible server).
  2. Paste your key. Petal stores it in the macOS keychain.
  3. Start from a preset (Clean Up, Email, Notes, Professional, AI Prompt, or Assistant) or write your own prompt.
  4. Insert variables such as {{name}}, {{app}}, and {{window}}, which Petal fills in on your Mac.
  5. Optional: turn on Date and Time, Web Search, or Screen. Screen sends a screenshot to the provider, needs Screen Recording access, and is off by default.
  6. Test the prompt in Try It.

You pay the provider for the tokens. Petal adds no charge.

Modes vs the per-app Router

Superwhisper modes are presets for a type of writing: Message, Email, Formal, Casual, Legal, Chat, or a Custom Mode with your prompt. You can link a mode to apps, and each mode has settings such as auto paste.

The Petal Router starts from where you are. In Settings > Router:

  1. Under Routes, add an app, or type a website such as github.com. A website route applies to any browser tab on that site or its subdomains.
  2. For each route, choose the Intelligence: Default, Off, Apple Intelligence, Petal W1, or Cloud Model.
  3. Choose a style: clean up, or paste as said.
  4. Edit the Instructions for that route.
  5. Under Speech Models, pick one model for dictation and a slower one for audio files.

"All Other Apps" is the fallback route. The result is the same idea as modes, but tied to the app or site, so you do not switch modes by hand. Read per-app dictation cleanup on Mac.

Clipboard handling

Both apps paste through the clipboard. A common problem in all such apps: the app puts your old clipboard back before a slow app reads the new text, so the old text pastes.

Superwhisper fixes (from its pasting docs):

  • Set Clipboard content to "Replace with dictation" if old text pastes.
  • Turn on Simulate keypresses to type instead of paste. It supports only US QWERTY.
  • For a double paste, turn off Simulate keypresses and do not press the shortcut during processing.

Read Superwhisper pastes old clipboard and Superwhisper not pasting for the full steps.

Petal waits for you to release modifier keys, sends Command-V, and restores your old clipboard after 700 ms. The delay is there because Electron apps such as VS Code and Slack read the clipboard late. If you do not want a restore, turn off Restore Clipboard in Settings > Recording. If a paste fails, use the Paste Last Transcription shortcut from Settings > General.

Setup time

Third-party reviews say that a full Superwhisper setup (models, permissions, and modes) takes 15 to 30 minutes. A basic setup is faster.

A basic Petal setup:

  1. Download Petal from GitHub Releases and move it to Applications.
  2. Allow the microphone.
  3. Turn on Petal in System Settings > Privacy & Security > Accessibility.
  4. Download Parakeet 110M (about 455 MB) in Settings > Models.
  5. Press Option-Space in a text field and speak.

The model download takes the most time. Router routes and cloud prompts are optional.

Other features

Feature Superwhisper Petal
Meeting recording Meetings, Mac only No
Translation to English Pro Through a cloud prompt
History Recent recordings Searchable history in ~/Documents/petal/history

Petal also has these settings, which this guide did not compare one by one with Superwhisper:

  • Send Now: while you record, press Control-X then Control-S to stop, paste, and press Return.
  • Trim Silence and Auto Speed-up in Settings > Recording.
  • Lower System Audio while you speak.
  • petal:// deep links and a Raycast extension.

License

Superwhisper is closed source. Petal is MIT licensed. You can read the code that handles your audio and your keys, build it yourself, and change it. The signed release is free and updates through Sparkle.

Which one to choose

Choose Superwhisper if you need Windows, iOS, or Android in one account, meeting recording, an Intel Mac, or modes that you already set up.

Choose Petal if you have an Apple silicon Mac on macOS 26.2 or later and want own-key cloud models, file transcription, and per-app routes for free, with open-source code.

Get Petal for Mac.

Questions

Are Petal and Superwhisper both offline?

Both can be. Superwhisper's free plan includes unlimited local Whisper models. Petal runs Parakeet, Whisper, Qwen3 ASR, and other models on the Mac. Both need a model download first.

Can I use my own OpenAI or Anthropic key for free?

In Petal, yes. Cloud Model in Settings > Intelligence takes your own OpenAI, Anthropic, or OpenRouter key, or a custom server such as Ollama or LM Studio, at no cost from Petal. In Superwhisper, your own API keys are a Pro feature.

What does Superwhisper Pro cost?

At the time of writing, Superwhisper lists Pro at $8.49 per month, $84.99 per year, or $249.99 lifetime, with 40% off for students and a 14-day refund. Check superwhisper.com for current prices.

Does Petal have modes like Superwhisper?

Petal uses a Router instead. Each app or website can have its own route with its own cleanup model and instructions. Petal also has prompt presets such as Email, Notes, and Professional for cloud models.

Why does Superwhisper paste my old clipboard?

The app restores the old clipboard before the target app reads the new text. In Superwhisper, set Clipboard content to Replace with dictation. Petal waits 700 ms before it restores the clipboard.

ComparisonsMac

Best Dictation App for Mac: 8 Apps Compared

The best dictation app for Mac depends on price, privacy, and offline use. Compare Petal, Apple Dictation, Wispr Flow, Superwhisper, and 4 more.

ComparisonsMac

Petal vs Apple Dictation on Mac

Petal vs Apple Dictation on Mac: compare the silence stop, model choice, Petal W1 cleanup, offline use on Apple silicon, history, and what Apple does well.