Key fact: Muse Glimmer is a 30B Apache-2.0 model for local agents, not the consumer Muse app.
How we got it: Meta research post, 10 August 2026, and the Hugging Face card meta-models/Muse-Glimmer-30B.
Checked:
Meta Muse Glimmer 30B Local Hardware Guide (VRAM, Quants, Agent Stacks)
By LLM Hardware team · Methodology ·
Key facts
- Model
- 30B, Apache-2.0, local. Distinct from the Muse app.
- Q4 formula
- 20 GB. Meta K-quant language model about 17 GB.
- Q8 formula
- 38 GB. Full precision: Meta says over 55 GB; formula sketch is 74 GB.
- Checked
- 2026-10-13
Checked 2026-10-13. Author: LLM Hardware team. This site does not store or show Amazon prices. Check current price on Amazon.
Citeable facts
Key fact: Site-formula Q4 for 30B is 20 GB. Meta’s own K-quant language model is about 17 GB and is aimed at a 24–32 GB envelope once KV, perception, and the drafter are included.
How we got it: 30 × 0.5 × 1.2 + 2 = 20. Meta’s post: 4-bit language-model weights under 20 GB, K-Quant-17GB called out by name, full precision over 55 GB.
Checked:
Key fact: Q8 for the same 30B is about 38 GB. That is a 48–64 GB unified-memory tier, not a 24 GB GPU.
How we got it: 30 × 1.0 × 1.2 + 2 = 38.
Checked:
Methodology — parameter math, quantization bytes, and the source list.
Muse Glimmer is Meta’s open-weight agent model: 30 billion parameters, Apache-2.0, built for tool use, failure recovery, screenshots, and local coding. Weights are on Hugging Face. This is not consumer Muse. That product stays on a cloud VM and is powered by Muse Spark. Buying a GPU does not install the Muse app, and installing the Muse app does not put Glimmer on your disk.
What Meta actually promised about memory
The 10 August 2026 research post says full precision for a 30B model is over 55 GB, which is already past any single consumer GPU. They compress the language-model weights to about 4-bit, under 20 GB, and they name a K-Quant around 17 GB. The working set they describe is that language model plus KV cache, a perception encoder for images, and a small DFlash drafter. Their envelope for all of that is 24 GB or 32 GB. They say they measured that 17 GB quant with the drafter on an M4 Max, an M5 Max, and an RTX 5090. The speeds are theirs. This page does not retype a speed table from that post, so it prints no tokens per second. The local-models guide quotes the model card’s speed table if you want those figures attributed.
Meta also said llama.cpp, MLX, and ExecuTorch integrations were landing, and that Ollama, LM Studio, and Unsloth were partner paths “in the coming days” from 10 August. By 2 October that may or may not be a one-click pull in your app. Confirm the tag exists before you plan a weekend around it. vLLM and SGLang are the multi-user serve path they named, which is a LAN box, not a laptop.
Quant table: their number and ours
Two numbers can both be honest. Meta measured a specific K-quant. Our VRAM formula is a planning estimate that does not know their exact quantization recipe: params × bytes × 1.2 + 2. Use the vendor file size when you have it, and the formula when you are comparing Glimmer to Qwen3.8-27B or K2-32B. Check a context length on the VRAM calculator.
| Mode | Figure | Box that matches |
|---|---|---|
| Q4, VRAM formula | 30 × 0.6 + 2 = 20 GB | 24 GB GPU, or 32 GB unified, before a long chat |
| Meta K-quant LM | About 17 GB language model. Under 20 GB at 4-bit. | 24–32 GB once KV, vision, and the drafter join |
| Q5 | 30 × 0.625 × 1.2 + 2 = 24.5 GB | 32 GB, tight. 36 GB unified is calmer. |
| Q8 | 30 × 1.2 + 2 = 38 GB | 48–64 GB unified |
| BF16 / FP16 | Formula 30 × 2.4 + 2 = 74 GB. Meta: over 55 GB at full precision. | Not a 24 GB or 32 GB card. Quote Meta’s 55 GB for their weights; use 74 GB only as the formula’s overhead-inclusive sketch. |
A 16 GB gaming card is below the target Meta published. Do not buy one “for Glimmer.” A 12 GB card is a 7B machine: K2-Horizon-7B at Q4 is 6.2 GB. Different job.
Agent stack around the weights
Glimmer is the brain. It still needs a scaffold. Meta says it works with OpenClaw and other orchestration patterns, and they show an OpenCode demo. This guide does not deep-link OpenClaw; the project URL was not re-confirmed in this pass. What this guide points to:
- llama.cpp, MLX, or ExecuTorch, which is the set Meta named for on-device runtimes.
- LM Studio or Ollama once the integration is actually in the build you downloaded.
- Open WebUI or OpenHands as the agent UI. OpenHands’ current docs point at Agent Canvas and the SDK. The old Local GUI is deprecated.
- vLLM or SGLang only if more than one person is hitting the model. That is a memory-bandwidth problem on top of the weight size.
Screenshot tasks use the perception encoder, so they sit in the 24–32 GB envelope Meta described, not in the 17 GB language-model file alone. Text-only tool calls can be leaner. Do not plan the vision agent on the language-model number.
What to buy
The comfortable single-user machine in this catalog is the Mac Studio M5 Max 36 GB, ASIN B0HGKSQMX6. Check current price on Amazon. A 24 GB discrete card also matches Meta’s lower envelope; see an RTX 4090 rather than a 16 GB Blackwell card. If you want Q8 (38 GB) or to serve Glimmer beside another model, 36 GB is the floor and 64 GB configured at Apple, or the EVO-X2 128 GB (B0F53MLYQ6), is the headroom. The 64 GB Studio was not a clean Amazon buy in that pass.
Mac Studio M5 Max 36 GB on Amazon
ASIN B0HGKSQMX6. Amazon Associates link. The live price is on that page, not locked in here.
ASIN B0BG94PS2F. Amazon Associates link. The live price is on that page, not locked in here.
Related guides
- VRAM calculator
- Agent workload formula, worked
- Qwen3.8 and K2 next to Glimmer
- Consumer Muse, which does not run here
Sources
- Meta research, Introducing Muse Glimmer (10 August 2026)
- meta-models/Muse-Glimmer-30B
- Meta developer model page
- Muse Glimmer deploy docs
A correction or a hardware question goes to the contact page.