Superwhisper Review 2026: Is It Worth $249 Lifetime?

Item: Superwhisper
Rating: 4.3
Author: Spokenly

Verdict at a Glance

4.3 / 5

Superwhisper covers the standard local model lineup (Whisper Tiny through Large V3 Turbo, Parakeet V2 and V3) with a customizable per-app modes system, sold either as $8.49/mo subscription or $249.99 lifetime. Setup is heavy for non-technical users, several privacy defaults are worth a second look (audio saved to disk by default, plaintext API keys), and the price is the highest in the category outside of Wispr Flow.

Better fit for most users: Spokenly is free with the same local Parakeet and Whisper stack, BYOK to OpenAI, Deepgram, and Groq at zero markup, and ships an MCP server for AI coding agents that Superwhisper does not have built in. Lower friction, lower cost, same offline guarantees.

What works

Largest local model catalog: Whisper Tiny through Large V3 Turbo, Parakeet V2 and V3
Cloud BYOK to Deepgram Nova 3, ElevenLabs Scribe V2, OpenAI
Custom modes with user-written prompts, per-app auto-activation
Context awareness: reads active app, clipboard, selected text

What does not

Power-user oriented, new users report feeling overwhelmed
Audio recordings saved to disk by default with no opt-out toggle (top feedback request)
API keys stored in plaintext JSON, not Keychain
Recording window overlay flagged as intrusive by some users

What Is Superwhisper

Superwhisper homepage at superwhisper.com showing the tagline 'Just speak. Write faster.' with Download for Mac and Download for Windows buttons over a dark blue gradient — Superwhisper homepage at superwhisper.com, May 2026.

Superwhisper is a system-wide AI dictation app built by SuperUltra, Inc., a bootstrapped Toronto company led by Neil Chudleigh (formerly co-founder of PartnerStack). It launched on Hacker News in August 2023 and has built a sizable user base in the Mac power-user community.

The product runs on macOS, Windows, and iOS. The default hotkey is Option-Space. You hold to speak, release to transcribe, and the text auto-pastes into whatever app is in focus. Custom Modes shape how each transcript is formatted (casual chat, email, code, custom prompts) and auto-activate based on the app you are using.

The headline pitch is "AI Voice to Text for macOS, Windows & iOS, write up to 5x faster using your voice." Privacy is the secondary pitch: local Whisper and Parakeet keep audio on-device when you pick those models (cloud BYOK is also supported). Spokenly offers the same architecture with a free tier instead of $8.49/mo or $249.99 lifetime.

Pricing in 2026

Superwhisper has three paid tiers plus a free tier. The lifetime option is the standout because it cuts off the subscription clock.

Tier	Price	Included
Free	$0	Small local Whisper models, 100+ languages, up to 3 custom modes, 15-minute Pro trial
Pro Monthly	$8.49/mo	All Whisper and Parakeet models, BYOK cloud, all custom modes, full feature set
Pro Annual	$84.99/yr	Same as Pro Monthly with annual discount
Lifetime	$249.99 one-time	Same as Pro, no recurring charges, covers Mac + Windows + iOS
Enterprise	Custom	SOC 2 Type II, team licensing

A 30-day no-questions-asked refund applies to all purchases. Renewals are not refundable. Third-party sources mention a 40% student discount via the Student App Centre, but it is not listed on Superwhisper's main pricing page.

Math: lifetime is worth it if you stay 2.5+ years. Wispr Flow Pro annual is $144/year, so $249.99 lifetime breaks even after about 21 months. Spokenly Pro is $9.99/mo or $99.99/yr, which is cheaper short-term but lacks the lifetime path. MacWhisper at €59 lifetime undercuts Superwhisper substantially if you only need file transcription.

Models: Local + Cloud BYOK

The model lineup is the most thorough in the category.

Local (on-device, Apple Silicon recommended)

Whisper Fast, Nano, Standard, Pro, Ultra V3, Ultra V3 Turbo (OpenAI Whisper family from tiny to large-v3-turbo)
Parakeet V2 (NVIDIA, English-only, fastest on M-series)
Parakeet V3 (NVIDIA, 25 languages, near-realtime)
On-device LLMs for post-processing: GPT OSS 20B, DeepSeek R1 Distill, Llama 3.x, Mistral 7B, Phi-2

Cloud (BYOK, proxied through Superwhisper)

Transcription: Ultra and S1-Voice (Superwhisper proprietary), Scribe V2 (ElevenLabs), Nova 3 / Nova 2 / Nova Medical (Deepgram)
Cloud LLM post-processing: OpenAI GPT-5, GPT-5.1, GPT-5.2 family
Anthropic Claude Sonnet 4.5 / 4.6, Haiku 4.5, Opus 4.6 / 4.7
Google Gemini 3 Flash and 3.1 Flash Lite
xAI Grok 4.1 Fast, Meta Llama 3.1 8B via Groq

Spokenly covers the same local stack (Parakeet V2/V3, Whisper Large V3 Turbo) and the same major cloud providers (OpenAI, Deepgram, Groq) for free with BYOK. Superwhisper goes deeper on model granularity and BYOK LLMs for post-processing. If you want hundreds of model permutations to tune, Superwhisper wins. If you want a curated stack that works out of the box, Spokenly wins.

Custom Modes

Custom Modes are Superwhisper's main customization layer. Each mode is a saved configuration: a hotkey, a transcription model, an optional LLM post-processing prompt, and auto-activation rules.

Practical examples from the community:

Slack mode: keep casual tone, no salutation, drop signatures
Email mode: formal greeting, signature, paragraph structure
Code mode: strip punctuation, lowercase, preserve technical terms
Terminal mode: paste shell commands without auto-formatting
Meeting mode: bullet point output, capture action items
Multilingual mode: auto-detect language, do not translate

Auto-activation switches modes based on the active app or website. You can also bind each mode to its own keyboard shortcut (added v2.12.0, April 2026). Spokenly covers the same per-app customization through custom AI prompts and word replacements without the upfront cost.

Platforms and System Requirements

macOS 13.3 or later. Apple Silicon strongly recommended for local models. Intel Macs run cloud models but local performance is poor.
Windows 10 or later, x64 and ARM64.
iOS 18 or later. Custom keyboard included in the same license.
No Android. No browser extension.
Single license covers Mac + Windows + iOS, including the lifetime tier.

The Windows build is newer (launched 2025) and less polished than the Mac build. Feedback board complaints include Windows crashes and target-app freezes. If Windows is your primary platform, test the 15-minute Pro trial before buying.

Privacy and Data Handling

Local mode is fully private: audio never leaves the device, no internet required. Cloud mode is proxied through Superwhisper's infrastructure so providers (OpenAI, Deepgram, ElevenLabs) never see your account, and the privacy policy states data is not retained for training.

Concerns worth knowing

Audio recordings save to disk by default with no opt-out toggle (23+ upvotes on the feedback board asking for one)
API keys store in plaintext JSON in ~/Library/Application Support, not Keychain (15+ upvotes asking for Keychain migration)
Closed source, so external audit relies on Superwhisper's word
SOC 2 Type II is Enterprise-tier only

For workflows that demand fully air-gapped operation, Spokenly's Local Only Mode blocks all outbound network requests including telemetry. Same local model stack, stricter network isolation guarantee.

Performance on Apple Silicon

Apple Silicon performance is where Superwhisper shines. Independent benchmarks:

M1: Whisper Large V3 runs at roughly 1x realtime
M4: Whisper Large V3 runs at 17 to 27x realtime
Accuracy stays 99.2-100% across model sizes on M4 (only latency changes)
Whisper Large V3 Turbo is the recommended default: 8x faster than Large V3 with WER difference of only 0.39 percentage points
Model sizes on disk: Large V3 is 2.9 GB, Large V3 Turbo is 1.6 GB

For 8 GB Macs, stick with Large V3 Turbo. 16 GB+ Macs can run any model. Parakeet V3 multilingual was added shortly after NVIDIA's release and runs near-realtime with built-in keyword recognition for up to 1,000 custom terms.

What Users Are Saying

Public ratings sit at 4.9/5 on Product Hunt (20 reviews) and 4.4/5 on Mac App Store (762 ratings). The Mac App Store sample is large enough to take seriously; the Product Hunt sample is small and skews toward early adopters who tend to rate generously.

Top praise points

Daily usability, speed, and accuracy on Apple Silicon
Custom modes with prompt layering (most-cited differentiator)
Privacy-first stance with 100% on-device option
Active development, near-monthly feature releases
Bootstrapped indie feel: support is responsive, no investor-driven feature pressure

Top complaint points

$249.99 lifetime feels steep relative to MacWhisper at €59 or Voibe at $149
Setup curve: new users report feeling overwhelmed by options
Audio saved to disk by default (top privacy feedback request)
Recording window overlay is intrusive for some workflows
iOS app is rated less reliable than desktop in App Store reviews

Superwhisper Alternatives

Free alternative

Spokenly

Same local Parakeet V3 and Whisper Large V3 Turbo stack. BYOK to OpenAI, Deepgram, and Groq at zero markup. MCP server for AI coding agents. Custom AI prompts. Agent Mode for macOS automation. Free with local + BYOK, Pro $9.99/mo. Lower setup friction than Superwhisper, no upfront $249.99.

Download Spokenly free

Wispr Flow

Cloud-only cross-platform with AI cleanup. Best for Windows and Android users. $15/mo, no offline.

See the comparison

MacWhisper

Best for file transcription (podcasts, meetings). €59 lifetime, Mac only, drag-and-drop with speaker diarization.

See the comparison

Apple Dictation

Free, built into macOS/iOS. Limited on accents and jargon, but on-device and zero setup.

See the comparison

Who Should Use It

Choose Superwhisper if

You are a Mac power user who wants every model option
You want a lifetime license to escape subscriptions
You dictate across many different apps with distinct formatting needs
Privacy matters and you want local-by-default with optional cloud BYOK
You enjoy tuning prompts and configuring per-app behavior

Skip Superwhisper if

$249.99 upfront is more than your budget allows
You want plug-and-play with minimal configuration
Android support is required
You need MCP-server-level integration with AI coding agents
File transcription is your main job (MacWhisper is cheaper for that)

FAQ

Is Superwhisper worth $249.99 lifetime?

If you plan to use it for more than 2.5 years and value owning vs renting, yes. The lifetime tier beats Wispr Flow's $144/year subscription after 21 months and covers Mac, Windows, and iOS with one purchase. For shorter horizons, the $8.49/mo monthly tier or the $84.99/year annual plan are cheaper entry points. If you want the same local model stack for free, Spokenly does it.

Does Superwhisper work offline?

Yes. Local Whisper models (Tiny through Large V3 Turbo) and Parakeet V2/V3 run entirely on-device on Apple Silicon Macs. No internet required for transcription. Cloud models (Deepgram Nova, ElevenLabs Scribe V2) and BYOK LLMs (GPT, Claude, Gemini) require connectivity, but you can opt for fully on-device mode in any custom mode.

Is Superwhisper better than Wispr Flow?

They serve different priorities. Superwhisper is local-first, customizable, and works offline. Wispr Flow is cloud-only, polished, and cross-platform with stronger AI cleanup. Superwhisper has a lifetime option ($249.99) that beats Wispr Flow's subscription after 21 months. Wispr Flow has team features and Android support that Superwhisper lacks. Privacy and offline use favor Superwhisper.

Is Superwhisper better than MacWhisper?

Different jobs. Superwhisper is for system-wide live dictation with per-app custom modes. MacWhisper is for file transcription (audio and video files, podcasts, meetings) with batch processing and speaker diarization. MacWhisper is significantly cheaper at €59 lifetime. Many users run both. If you only need live dictation, Superwhisper or Spokenly. If you only need file transcription, MacWhisper.

What is Superwhisper's Custom Mode?

Custom Modes are user-defined prompts that change how Superwhisper formats the transcript. Each mode has its own prompt, model selection, hotkey, and auto-activation rules. Example: a Slack mode that keeps the tone casual, an Email mode that formats with greeting and signature, a Code mode that strips punctuation. Modes can switch automatically based on the active app.

Is Superwhisper safe to use?

For local mode, yes: audio never leaves the device. For cloud mode, requests are proxied through Superwhisper so providers like OpenAI or Deepgram never see your account, and nothing is retained for training. Concerns from the user feedback board: audio recordings save to disk by default with no opt-out toggle, and API keys are stored in plaintext JSON in Application Support rather than Keychain.

Can Superwhisper work with Claude Code?

Yes. Superwhisper added Claude Code and Open Code agent integration in v2.13.0 (April 2026, macOS only). The integration pipes transcripts into the terminal agent. For full MCP-server integration where AI agents pull voice on demand, Spokenly's built-in MCP server is more direct.

Does Superwhisper support multiple languages?

Yes. 100+ languages and dialects with automatic language detection via Whisper. Parakeet V3 multilingual covers 25 languages locally on Apple Silicon. Translation to English is supported in transcription mode. UI is in English.