Muse Glimmer 30B
Meta Muse Glimmer 30B is an Apache 2.0 open-weight, dense multimodal agentic model designed for always-on local workflows. It combines text and image understanding, long-context reasoning, coding, tool calling, and computer-use capabilities. Meta publishes BF16 weights, official GGUF quantizations targeting 24GB and 32GB VRAM, and ExecuTorch packages for Metal and CUDA.
Deployment and license note
Muse Glimmer understands images but does not generate images. Meta targets 24GB VRAM with the official 17GB K-Quant build, 32GB VRAM with the higher-quality dynamic K-Quant build, and 64GB VRAM for full precision. Long contexts and vision inputs add runtime memory, so leave extra headroom. DFlash acceleration is optional and depends on runtime support.
Task Fit
Tool use, repo work, terminal workflows, and coding benchmarks.
Code generation, debugging, refactoring, and benchmark signal.
General writing, Q&A, and assistant use.
Document QA benefits from long context and instruction following.
Image or visual understanding, not necessarily image generation.
Not marked for image generation in the current library.
Not marked for video generation in the current library.
Not marked for voice in the current library.
Source Confidence
Variants and Quant Artifacts
Choose the artifact first; hardware fit follows from RAM, VRAM, format, and runtime.
| Quant | Format | Quality | Min RAM | Reco RAM | Runtime | Action |
|---|---|---|---|---|---|---|
| Q4_K_M | gguf | balanced | 24GB | 32GB | llama.cpp, lm-studio, mlx, executorch | Plan with this |
| Q4_K_XL | gguf | balanced | 32GB | 40GB | llama.cpp, lm-studio, mlx, executorch | Plan with this |
| BF16 | safetensors | high | 64GB | 80GB | transformers, vllm, sglang, executorch | Plan with this |
Recommended Hardware
Lowest estimated 5-year cost that can run this model.
Plenty of fast-memory headroom for this model.
Highest local performance signal among compatible hardware.
Benchmarks
Source and Review
Execution evidence
Run Muse Glimmer 30B with a documented recipe
Recipes connect hardware, a model artifact, tools, settings, verification, and a reportable result.
No verified recipe is linked to this record yet.
Compatibility estimates remain available in the planner. A recipe appears here only after its exact stack and verification protocol are documented.
Similar Models
DeepSeek V4.1 Flash is DeepSeek's MIT-licensed open-weight multimodal MoE for reasoning, coding, agents, and native image understanding. The complete checkpoint is approximately 763B parameters, including a 552B backbone; its Causal Encoder-Decoder architecture activates 8B parameters during input prefill and 16B during output decoding. CSA2 and FP4 KV caching reduce long-context cache cost. The model supports a 1M-token context window and is available through official FP8 weights, the DeepSeek API, and Ollama Cloud. It remains a data-center-class deployment despite the low active-parameter count.
Moonshot AI's Kimi K2.7 Code. Auto-imported from Hugging Face seed; review memory and benchmark details before using for final ranking.
Moonshot AI's Kimi K2.6. Auto-imported from Hugging Face seed; review memory and benchmark details before using for final ranking.