> ## Documentation Index
> Fetch the complete documentation index at: https://docs.stenoai.co/llms.txt
> Use this file to discover all available pages before exploring further.

# Summarization models

> Steno generates meeting summaries using local Ollama models by default, with optional cloud model support. Choose a model based on your hardware, privacy requirements, and the quality of notes you need.

After transcription, Steno passes the transcript to a language model to generate structured notes: a summary paragraph, key topics, and action items. By default, this uses a local model running via [Ollama](https://ollama.com) entirely on your Mac. You can optionally configure a cloud model (OpenAI, Anthropic, or custom endpoint) — see the section below.

## Available models

| Model             | Size    | Speed    | Quality   | Best for                                |
| ----------------- | ------- | -------- | --------- | --------------------------------------- |
| `llama3.2:3b`     | \~2GB   | Fast     | Good      | Quick meetings, daily use **(default)** |
| `gemma3:4b`       | \~2.5GB | Fast     | Good      | Lightweight alternative to llama        |
| `qwen3.5:9b`      | \~6.6GB | Moderate | Very good | Structured output, action items         |
| `deepseek-r1:14b` | \~9GB   | Slower   | Excellent | Complex meetings, detailed analysis     |
| `gpt-oss:20b`     | \~14GB  | Slow     | Excellent | Highest quality local output            |

Larger models produce better-structured, more detailed notes. The trade-off is processing time and disk space.

## How to change models

1. Open **Settings → Models → Summarization**
2. Select a model from the list
3. If the model is not yet downloaded, Steno will download it via Ollama

The new model is used for all future recordings. You can re-summarize an existing recording from its detail view using any model.

## Choosing the right model

**For daily use:** `llama3.2:3b` (default) works well for most meetings. Notes are clear and concise without requiring significant processing time.

**For important meetings:** `qwen3.5:9b` or `deepseek-r1:14b` produce more detailed notes with better-identified action items and more accurate key topics.

**For hardware with limited RAM:** Stay with `llama3.2:3b` or `gemma3:4b`. The larger models require at least 16GB RAM to run comfortably without swapping.

## Using a cloud model

Steno optionally supports OpenAI, Anthropic, or a custom API endpoint as an alternative to a local model. If configured, your transcript (not your audio) is sent to that API for summarization.

To configure a cloud model:

1. Open **Settings → Models → Summarization**
2. Select **Cloud model**
3. Choose the provider and enter your API key

Cloud models are off by default. If you work with confidential recordings, use a local model.

***

<Accordion title="Which model produces the best action items?">
  `qwen3.5:9b` is particularly good at identifying and formatting action items from meeting transcripts. If structured output is important, it is worth the extra size over the default `llama3.2:3b`.
</Accordion>

<Accordion title="How much disk space do the models require?">
  The default setup (`whisper-small` + `llama3.2:3b`) requires approximately 2.5GB. The largest model (`gpt-oss:20b`) requires \~14GB. Models are stored in `~/.ollama/models/`.
</Accordion>

<Accordion title="Can I delete models I no longer use?">
  Yes. Run `ollama rm [model-name]` in Terminal to remove a model and free the disk space. You can re-download it later from within Steno.
</Accordion>

<Accordion title="My Mac has 8GB RAM -- which model should I use?">
  Stick with `llama3.2:3b` (the default). It runs comfortably on 8GB. The 9B and larger models will cause swapping on 8GB, significantly slowing both summarization and your Mac.
</Accordion>