Skip to main content
Murmint uses two models. They can come from different providers.

Transcription

Turns your voice into words. Murmint 1, Murmint’s own speech model, runs on your PC for free and is the default.

Formatting

Cleans up and formats what you said, using a provider you add. A small, fast model is plenty.

Speech and formatting side by side

The top of the page has a panel for each job, with the model it uses right now. Pick a model opens a list grouped by provider. Choose one, and add backups under If it fails, try. If Murmint 1 is not on your PC yet, the Speech panel has a button to download it.
The Models page with the Speech to text and Formatting panelsThe Models page with the Speech to text and Formatting panels

Your providers

Your providers are shown as tiles under Your providers. Formatting uses these. Click a provider to add or change its key. Murmint saves the key, tests it, and tells you straight away if it works.
1

Click Add a provider

Pick one from the grid: OpenRouter, Groq, Gemini, Mistral, xAI, DeepSeek, Ollama, LM Studio, or any OpenAI-compatible server.
2

Paste your key

Local ones like Ollama and LM Studio need no key.
3

Choose the models you want

Murmint lists what the provider offers.
Your providers on the Models page
You can remove any provider, even the built-in ones. They come back from Add a provider whenever you want them.
OpenAI models need a key from platform.openai.com with its own billing. A ChatGPT Plus or Pro subscription does not include one.