# Run Muse Glimmer locally

> Source: <https://lmstudio.ai/blog/muse-glimmer>
> Published: 2026-08-10 00:00:00+00:00

# Run Muse Glimmer locally

**We partnered with Meta to bring launch day support for Muse Glimmer in LM Studio Bionic!**

[Muse Glimmer](/models/muse-glimmer) is a new 30B open-source model from Meta that excels in agentic tasks, and can fit right on your laptop. The model is available starting today for downloading and running locally in LM Studio Bionic. Muse Glimmer is available under the Apache 2.0 license.

## Run Muse Glimmer locally with Bionic

- Download and install Bionic
- Go to Settings → Explore to find Muse Glimmer from the catalog
- Select the model and click Download

Once the download finishes, Muse Glimmer will appear as a local model option for any existing or new session.

## Muse Glimmer for local agentic work

Muse Glimmer is a 30B-parameter open-source model optimized for local agent workflows. It supports vision input and demonstrates strong tool use and multi-step task execution for complex agentic tasks.

The model is a great fit for working with the Bionic agent locally. Bionic's friendly interface makes it suitable for both technical and non-technical users, while Muse Glimmer is optimized for running locally on both Mac and PC machines. Pairing the two together means you can run agentic workloads on your own hardware 24/7, reducing worries about data leakage or token costs.

## BionicBench v0.1

To understand how well local models handle end-to-end agentic work, we're developing BionicBench: an internal evaluation of real-world workflows in Bionic. Its first iteration includes 18 tasks spanning coding and following repository instructions, handling text files and attachments, understanding images and screenshots, editing Word documents, PowerPoint decks, and Excel spreadsheets, and generating PDFs. The cases test whether an agent can follow detailed instructions, create or modify the requested artifact, and preserve formatting and file structure where required.

Using the latest graded outcome for each task under the same local test setup, Muse Glimmer completed 88.9% of tasks. This was ahead of Gemma 4 31B at 83.3% and Qwen 3.6 27B at 72.2%, demonstrating strong breadth for a model that runs on your own hardware.

BionicBench v0.1 results for Muse Glimmer

Here are some ways to use Muse Glimmer with Bionic:

### 1. Deep research agent, right on your Mac or PC

Research a topic with Bionic and preview the PDF summary right in the app

Ask Bionic to research a topic and leverage its native web search functionality to gather current information from the web. Point Bionic to a local folder or drag in files to give Bionic context, then ask it to summarize documents or produce detailed research reports. Bionic's agentic document understanding allows you to work seamlessly across large numbers of files and a variety of sources.

### 2. Use Muse Glimmer to convert rough outlines into polished presentations

Use Muse Glimmer in Bionic as a thinking partner and ask it to help you shape thoughts into a clear outline. Then, ask Bionic to turn that outline into a polished deck right in the same session. Iterate and refine your work by asking the agent for changes, and easily rollback changes to previous versions with automatic checkpointing.

### 3. Run multi-step coding workflows

Toggle on Allow Coding when creating a project to let Bionic run commands in a local directory

Create a new project in Bionic and toggle on Allow Coding to use Muse Glimmer with a fully local coding agent. Bionic is great at multi-step coding workflows, such as updating release notes on recent git commits, refactoring across files, or fixing failing tests. None of your data goes to third party providers and you pay no per token inference cost.

### 4. Automate everything with the Bionic local API and Muse Glimmer

#### From the terminal

```
lms server start
```

The local API conforms to OpenAI's Chat Completions. This means you can point a wide variety of tools and frameworks to it seamlessly.

#### From the app

Under Settings → Local Model API, toggle on the Local API server to point your projects at Bionic and Muse Glimmer. You can use this with automation scripts, or wire it into batch jobs and CI to have a capable agent running entirely on your hardware.

## Learn More

Download [Muse Glimmer 30B](/models/muse-glimmer) and run it locally.

Check out [our docs](https://lmstudio.ai/docs) to learn more about using local models in Bionic.
