Superwhisper

Superwhisper is a voice-to-text tool built by Superultra Inc. that transcribes speech and uses AI to refine the output directly inside any app — from Slack and Gmail to Cursor and Notion. It runs on Mac, Windows, and iOS, and works both online and offline. The homepage states that "hundreds of thousands" rely on it to save time, and the product has attracted public endorsements from notable figures in the tech community.

What It Is

Superwhisper sits in the speech-recognition-plus-AI-polish category. Rather than just dumping raw transcription into a text field, it applies configurable AI modes — Formal, Casual, Legal, Chat, and custom-built modes — to shape tone, structure, and formatting before the text lands in whatever app you're using. The result is pasted automatically via clipboard integration, so there's no manual copy-paste step. It supports over 100 languages and dialects and can translate any of them to English.

Core Feature Set

• Offline transcription — works without Wi-Fi using on-device models (Apple Silicon Macs perform best; Intel Macs are recommended to use cloud models)
• Predefined and custom modes — preset modes for messages, emails, voice notes, and more; custom modes let users define formatting rules, structure preferences, and specialized prompts
• Multi-model AI backend — users can choose from GPT-5, Claude Haiku 4.5, Llama 4, Grok 4.1, Gemini 3.0 Flash, Ministral, and others; Pro users can supply their own API keys
• Custom vocabulary — enter names, abbreviations, and specialized terms once; Superwhisper remembers them permanently
• Meeting assistant — records and digests meetings with automatic notes
• File transcription — upload audio or video files for reliable transcripts
• Push to talk — hold, speak, release control scheme
• Super Mode — AI-enhanced mode that adapts to screen context for smarter results
• Custom keyboard shortcuts — launch, dictate, and control via configurable hotkeys

Platform and Integration Reach

Superwhisper is available as a native desktop app for Mac and Windows, and as an iOS app. It integrates with any app that accepts text input — the homepage specifically calls out Slack, Gmail, Cursor, Notion, Telegram, and WhatsApp. The product also highlights a dedicated workflow for agentic coding tools: it lists Claude Code, Opencode, Amp, Codex, and other AI coding agents as supported environments, positioning voice input as a way to give agents more context without typing.

Audience and Adoption Signals

The homepage displays testimonials from Andrej Karpathy (Eureka Labs), Pieter Levels, Andrew Wilkinson (Tiny), and Guillermo Rauch (Vercel), all published on the Superwhisper website. Karpathy's quote specifically connects Superwhisper to the "vibe coding" workflow he described publicly. The vendor states "hundreds of thousands" of users rely on the product, though this figure comes from the vendor's own homepage copy. Company logos shown on the homepage include Vercel, Spotify, OpenAI, Shopify, Meta, and Apple, though the site does not explicitly state these are customers.

Deployment Model and Offline Capability

Superwhisper runs as a native desktop application rather than a browser extension or web app, which enables its offline transcription capability. Offline models run best on Apple Silicon Macs; Intel Macs and Windows users are directed toward cloud models. The iOS app extends the same voice-to-polished-text workflow to mobile. Pro users can bring their own AI API keys, giving them control over which models process their audio and where data is sent.

Current Status

The product is actively maintained with a public changelog at superwhisper.com/changelog and a public roadmap. The homepage references GPT-5, Claude Haiku 4.5, Llama 4, Grok 4.1, and Gemini 3.0 Flash as available models, indicating the model list is kept current. The company states it is hiring and is backed by API Capital, as noted in the footer of the pricing section. A 30-day no-questions-asked refund is offered on all paid plans.