# Local mode: offline by default. Dimmy Help.

> Local mode: offline by default. Step-by-step help for Dimmy, the native voice dictation app for macOS, Windows, and Linux.

Source: https://dimmy.app/help/local-mode/

[← Back to Help center](https://dimmy.app/help/)

Providers & Models

# Local mode: transcription on your own machine

Four speech engines that run on your device. No API key, no upload, no per-minute cost. Works on a plane.

info

Local mode is the shipped default on macOS. On Windows and Linux the shipped default is cloud, and onboarding preselects Local for you unless Dimmy reads your graphics as too weak, in which case it preselects Cloud and shows you why. Either way you can change it in Settings → Voice input → Mode.

In local mode the recording never leaves the machine. On first launch Dimmy downloads a model: around 78 MB for Whisper base-q8\_0 on Windows, or the Parakeet bundle of about 466 MB on macOS, where Parakeet is the default engine. After that, transcription is free and works with no connection.

## What runs locally

* Speech to text, with four engines to choose between: whisper.cpp, Parakeet TDT v3, Qwen3-ASR and Whistle. See [Choose a local model](https://dimmy.app/help/whisper-models/).
* Filler removal (no model, just regex).
* LLM rewriting, optional, if you enable the on-device model. See [Optional on-device LLM](https://dimmy.app/help/local-llm/).
* Everything else: history, settings, key storage.

## Where it runs on each platform

| Platform             | Speech runs on                                                                                                                                   | Engines available            |
| -------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------ | ---------------------------- |
| macOS, Apple Silicon | Metal on the GPU, plus the Neural Engine for Parakeet, for Qwen's Neural Engine variants, and for Whisper once you download its Core ML encoder. | Whisper, Parakeet, Qwen3-ASR |
| Windows              | Vulkan on the GPU, falling back to the CPU on its own when the driver reports no usable device.                                                  | Whisper, Parakeet, Qwen3-ASR |
| Linux, AppImage      | CPU only. The AppImage stays CPU-only so it runs across distributions with different Vulkan loaders.                                             | Whisper                      |

info

There is no CUDA build of Dimmy. GPU acceleration on Windows is Vulkan, which works on any card whose driver exposes it, not only NVIDIA.

## Performance

Measured on a 4 GB NVIDIA T600 over Vulkan, same 15-second chunk in the same session: Whisper Base 1.5 s, Whisper Small 1.7 s, Whisper Large-v3-Turbo Q8 5.9 s, Qwen3-ASR 1.7B 2.6 s. Parakeet is the odd one out because its cost tracks the audio you actually give it: measured on the CPU alone, warm latency was 337 to 547 ms.

image · placeholder

Local benchmark chart: latency by hardware and clip length

info

Those are one machine and one recording. The same card measured about 8 seconds per window one morning and about 2 seconds the same afternoon, because its power limit was stuck at 20 W of 35\. Treat any number here as an order of magnitude, not a promise.

## When to use cloud instead

* You are on integrated graphics, where local models load but crawl.
* You record long sessions and want them back in seconds without tying up the GPU.
* You want realtime typing at the lowest possible latency. That path uses OpenAI Realtime or Deepgram when you have saved a key for either.

tip

Accuracy alone is no longer a reason to go to the cloud. On conversational Italian, local Qwen3-ASR 1.7B beat Whisper Large-v3-Turbo in our own measurements, including one sentence Whisper transcribed as the exact opposite of what was said. To work out what suits your machine, see [Which model should you use?](https://dimmy.app/help/choose-your-model/).

Was this helpful? 

## In this category

[ Providers & ModelsWhich model should you use?Four questions decide it: what machine you have, what must not leave it, which subscriptions you already pay for, and which language you speak. Answer them and the rest follows.](https://dimmy.app/help/choose-your-model/) [ Providers & ModelsCloud providers comparedGroq is fastest, OpenAI is the gold standard, Deepgram leads on streaming. All optional. Bring your own key.](https://dimmy.app/help/cloud-providers/) [ Providers & ModelsAdd your API keyOpen Settings → Providers & keys, find the provider card, paste the key. It is encrypted on your machine. You can update keys at any time.](https://dimmy.app/help/api-keys/) [ Providers & ModelsChoose a local modelThe whole on-device catalogue: nine Whisper sizes, Parakeet TDT v3, four Qwen3-ASR variants and Whistle. What each one weighs and what it runs on.](https://dimmy.app/help/whisper-models/) [ Providers & ModelsOptional on-device LLMRun the LLM enhancement pass locally via llama.cpp. No cloud, no key, no per-token cost. Pick a model to match your GPU memory.](https://dimmy.app/help/local-llm/)

Last updated: 2026-05-22

On this page

1. [01 What runs locally](#what-runs-locally)
2. [02 Where it runs on each platform](#where-it-runs-on-each-platform)
3. [03 Performance](#performance)
4. [04 When to use cloud instead](#when-to-use-cloud-instead)