Muse Glimmer 30B
Meta’s first open-weight model since Llama 4: a dense 30B multimodal reasoning model under Apache 2.0, built for local agents and coding on a single 24 GB GPU or 32 GB Mac.
What do you want to do?
Choose a task. Get the software, files, settings, and steps to run it.
Chat privately with Muse Glimmer 30B
Run Meta’s Apache 2.0 Muse Glimmer 30B with llama.cpp on Windows, Mac or Linux and chat with it, including about images, in your browser with every prompt staying on your computer.
Build a local AI agent with Muse Glimmer 30B
Run Meta’s Muse Glimmer 30B with llama.cpp, connect Hermes Agent using the settings on Meta’s GGUF card, and have it read two files and write a checked review.
Code locally with Muse Glimmer 30B in VS Code + Cline
Serve Meta’s Muse Glimmer 30B with llama.cpp, connect VS Code and Cline, and have it write a small module plus tests that you review and run yourself.
Open a guide for its software, hardware requirements, and setup steps.
At a glance
- Parameters
- About 29.6B, including a 1.8B vision encoder
- Architecture
- Dense multimodal transformer · text and image in, text out
- Context length
- 131,072 tokens
- License
- Apache 2.0 · commercial use allowed, no user cap
- Software
- llama.cpp, Ollama, LM Studio, Unsloth, vLLM, SGLang, Transformers
Will it run on your machine?
Save your machine to see a personalized rating and its reasoning.
Add my hardwareBest for
- Run a local Hermes agent that reads and writes files with your approval.
- Connect VS Code + Cline for private coding help.
- Private chat in your browser, including questions about images.
Tradeoffs
- Always reasons before answering; the reasoning channel cannot be switched off, only shortened.
- Dense model: partial offload to system RAM is much slower than with mixture-of-experts models of similar size.
- Benchmark claims against Gemma4-31B and Qwen3.6-27B are Meta’s own.
- Needs llama.cpp build b10353 or newer.
Ways to run it
llama.cpp · Windows, macOS, Linux
Meta’s own GGUF. Used by the chat, Cline and Hermes guides on this page.
llama-server -hf meta-models/Muse-Glimmer-30B-GGUF:Q4_K_M --jinja -c 32768 --temp 1.0 --top-p 0.95 --top-k 64Ollama · Windows, macOS, Linux
Listed on Meta’s GGUF card; not used by the guides here.
ollama run hf.co/meta-models/Muse-Glimmer-30B-GGUF:Q4_K_MUnsloth Studio · Windows, macOS, Linux
Unsloth’s desktop app and dynamic GGUFs; Unsloth puts the entry point at 18 GB of combined RAM and VRAM.
Official instructionsvLLM / SGLang · Linux GPU server
Full-precision serving from the base repository.
Official instructionsWhat to expect
Start with Meta’s Q4_K_M GGUF in llama.cpp. On a 24 GB GPU or a 32 GB Mac it is a strong, permissively licensed local agent and coding model.
Explore its uses
Sources and files
Primary sources and files behind this model page.
Keep exploring
GLM 5.3
Explore 2 versions of GLM 5.3. Compare model sizes, packages, hardware requirements, and available setups.
Kimi K3
An open-weight multimodal model for long-running coding, reasoning, and knowledge work.
Gemma 4
Explore 6 versions of Gemma 4. Compare model sizes, packages, hardware requirements, and available setups.