Meta's launch post for Muse, a personal agent powered by Muse Spark 1.3 that runs in a dedicated Muse Secure VM and acts across email, travel, forms and purchases.
Resource · Errands & personal agent
Your product
Sponsored
Put your logo, a line of copy and an image right here, between the builds Muse developers come to read. Same size as a post.
Put your product here
Shown every 12 builds · on every catalog page
turboderp's self-calibrated EXL3 quants of Muse Glimmer 30B down to 1.75 bits per weight, with calibration and eval traces.
Resource · Local & open models· ♥ 14
EXL3 quants of Muse Glimmer
MindStudio notes Muse Spark 1.3 topped DeepSWE at 75.4 and placed third on Artificial Analysis, yet in a hands-on game-clone test produced "a cube shooting at other cubes."
Resource · Benchmarks & research
Why Muse Spark 1.3's benchmark scores don't add up
GeekWire reports Amazon blocked Muse from making purchases as of September 20, saying the agent accessed its store without permission, didn't identify itself, and appeared to store credentials.
Resource · Benchmarks & research
Amazon blocks Meta's Muse shopping agent
DSpark drafter warm-started from Meta's DFlash assistant, adding a rank-256 Markov head and a confidence head (about 103M new parameters).
Resource · Local & open models· ♥ 8
DSpark drafter for Muse Glimmer
AdaptlyPost walks through connecting its scheduler to Muse as a custom connector: give Muse the public OpenAPI spec and docs, then paste a bearer token into Muse's secure credential prompt.
Resource · Connectors & MCP
Building a Muse connector for social posting
Your product
Sponsored
Put your logo, a line of copy and an image right here, between the builds Muse developers come to read. Same size as a post.
Put your product here
Shown every 12 builds · on every catalog page
MarkTechPost walked through Muse for Mac, which pulls context from Calendar, Mail, Messages, Notes and files for multi-step tasks, with opt-in access, optional Full Disk Access and approval before deleting files or sending messages.
Resource · Errands & personal agent
Muse for Mac: cross-app tasks explained
Solo-built GGUF line of Muse Glimmer 30B with custom calibration, per-tensor allocations and an eval harness behind every reported number.
Resource · Local & open models· ♥ 21
Muse Glimmer GGUF (AK line)
Decensored Muse Glimmer 30B made with Heretic v1.4.0, shipped with a reproduce directory.
Resource · Local & open models· ♥ 15
Muse Glimmer 30B heretic
A suite of 27 GGUF quantizations of an abliterated Muse Glimmer 30B produced with Heretic v1.4.0.
Resource · Local & open models· ♥ 20
Heretic Uncensored Muse Glimmer GGUF
mlx-community's 4-bit MLX conversion of Muse Glimmer 30B made with mlx-vlm 0.6.12 for Apple Silicon.
Resource · Local & open models· ♥ 18
MLX 4-bit Muse Glimmer
Meta's launch post for Muse Glimmer, an Apache 2.0 30B model for local agents that fits in ~20GB at 4-bit and runs on M4/M5 Max Macs, RTX 5090s or 24–32GB GPUs.
Resource · Local & open models
Introducing Muse Glimmer
Sebastian Raschka breaks down Glimmer's dense architecture: 3:1 sliding-window to global attention, 32 query heads with only 2 KV heads, and ~52 KiB of KV cache per token.
Resource · Benchmarks & research
Muse Glimmer 30B architecture notes
Resource · Local & open models· ♥ 4
Inferact NVFP4 W4A4 Muse Glimmer
Red Hat AI's FP8 weight and activation quantized Muse Glimmer 30B (text and image input) for vLLM.
Resource · Local & open models· ♥ 12
Red Hat AI FP8-block Muse Glimmer
Julian Goldie's GoldieBench scores Muse Spark 1.2 at 7.55/10 across 50 one-shot builds. It did well on fractals (8.7), a macOS desktop sim (8.6) and a galaxy viz (8.6), but 3D games like Doom and Dragonrealm often came out incomplete or black.
Resource · Games & 3D
GoldieBench: 50 one-shot demos of Muse Spark 1.2
Fireworks made Muse Glimmer 30B available on launch day, pitching high-concurrency, cost-effective serving for always-on agents.
Resource · Local & open models
Muse Glimmer 30B on Fireworks AI
GGUF quantizations of the Inco AI DFlash2 drafter for Muse Glimmer 30B, for use with the BeeLlama.cpp llama.cpp fork.
Resource · Local & open models· ♥ 1
DFlash2 GGUF for BeeLlama.cpp
Your product
Sponsored
Put your logo, a line of copy and an image right here, between the builds Muse developers come to read. Same size as a post.
Put your product here
Shown every 12 builds · on every catalog page
INT21 ran Muse Spark 1.1 in its SwarmOS orchestration layer and argues the durable enterprise advantage is orchestration, not any single model.
Resource · Agents & automation
Running Muse Spark 1.1 inside SwarmOS
TechCrunch's Connect roundup: retail connectors (Walmart, Best Buy, Gap, Sephora, Wayfair), Shop Pay and PayPal checkout, GitHub/Granola/Notion connectors, 1,500+ developer applications in a week, and Zuckerberg's plan to profit from a small fee on transactions.
Resource · Connectors & MCP
Everything new in Muse from Meta Connect
ROCmFP4 GGUF of Muse Glimmer 30B with DFlash for AMD Strix Halo, requiring a ROCmFPX llama.cpp fork that adds the muse-glimmer architecture.
Resource · Local & open models· ♥ 4
ROCmFP4 Strix Halo DFlash GGUF
Kingy AI explains Muse Glimmer's benchmarks, 24-64 GB hardware needs, GGUF setup, pricing, runtimes and risks.
Resource · Local & open models
Muse Glimmer 30B: benchmarks, hardware and how to run
KGP Talkie asked Qwen 3.8 27B, Muse Glimmer 30B and Gemma 4 26B the same twelve questions ten times each to see which answers the same way twice.
Resource · Benchmarks & research
Local model consistency test: Muse Glimmer vs Qwen vs Gemma
Meta's lightweight DFlash block-diffusion drafter for Muse Glimmer 30B that predicts blocks of 16 tokens per forward pass for speculative decoding.
Resource · Local & open models· ♥ 61
Muse Glimmer DFlash drafter (official)
Pre-exported ExecuTorch PTE artifacts of Muse Glimmer 30B from Meta, lowered and optimized for specific target backends.
Resource · Local & open models· ♥ 34
Muse Glimmer 30B ExecuTorch PTEs
ROCmFP4 and ROCmFP8 builds of Muse Glimmer 30B and its drafter, targeted and tested on AMD Strix Halo (gfx1151).
Resource · Local & open models· ♥ 18
Muse Glimmer ROCmFPX GGUF for Strix Halo
Simon Willison reads the Muse Code launch as more evidence that long-sequence agentic tool calling is now the key model trait.
Resource · Coding & dev tools
Simon Willison on Muse Code and Muse Spark 1.2
llama.cpp imatrix quantizations of Muse Glimmer 30B with image support via mmproj and MTP/DFlash notes.
Resource · Local & open models· ♥ 19
bartowski imatrix GGUF of Muse Glimmer
DeepLearning.AI's The Batch covers Muse's security design (isolated VMs, the Sentinel credential layer, prompt-injection classifiers) and its free tier of up to 100M tokens a week.
Resource · Benchmarks & research
How to secure agents for the masses (The Batch)
Meta's developer post on running Muse Glimmer on a single consumer GPU with vLLM, llama.cpp and ExecuTorch, with quantized builds in 24-32 GB and cookbook recipes.
Resource · Local & open models
Build with Muse Glimmer: local agents on one GPU
The smallest and fastest MLX 4-bit (group size 64) Muse Glimmer checkpoint, served with SGLang's MLX backend on Macs with 48 GB+ unified memory.
Resource · Local & open models· ♥ 11
Muse Glimmer MLX q4 for SGLang
Unsloth Dynamic 2.0 GGUF quants of Muse Glimmer 30B with a companion run guide and thinking toggles.
Resource · Local & open models★ Pick· ♥ 548
Unsloth Dynamic 2.0 GGUF of Muse Glimmer
LM Studio launched Muse Glimmer support, reporting it completed 83.3% of tasks on its 18-task BionicBench v0.1 versus 77.7% for Gemma 4 31B and Qwen 3.6 27B.
Resource · Local & open models
LM Studio: Run Muse Glimmer locally
Your product
Sponsored
Put your logo, a line of copy and an image right here, between the builds Muse developers come to read. Same size as a post.
Put your product here
Shown every 12 builds · on every catalog page
Resource · Local & open models· ♥ 18
Red Hat AI NVFP4 Muse Glimmer
Resource · Local & open models· ♥ 14
Unsloth bnb 4-bit Muse Glimmer
Simon Willison ran an 18.16GB build of Muse Glimmer locally, testing code exploration and image description. He found multi-step reasoning and tool use strong and creative generation mixed.
Resource · Local & open models
Simon Willison: Introducing Muse Glimmer
QAInsights' hands-on first look at Muse Code covering CLI commands, syntax and setup on launch day.
Resource · Coding & dev tools
Getting started with Muse Code CLI commands
AgentMail argues Gmail was built for a person at a keyboard, not an agent on a cloud VM at 5am, and makes the case for a separate agent inbox.
Resource · Connectors & MCP
Should you connect Meta Muse to Gmail?
Federico Ramirez uses Muse Code and Spark 1.3 for GitHub issue-to-PR workflows. He finds output readable and $15/month good for 2-4 hours of continuous coding, but skill execution inconsistent and the sandbox too restrictive.
Resource · Coding & dev tools
Muse Spark 1.3: a developer review
QUASAR-trained Blackwell W4A4 build of Muse Glimmer 30B at 21.8 GiB, reporting 20% lower KL to BF16 than Red Hat's NVFP4 W4A4 checkpoint.
Resource · Local & open models· ♥ 3
QUASAR native NVFP4 Muse Glimmer
137 builds · page 1 of 4
About this shelf
Resources are reference material rather than builds, and there are 137 of them. They include Meta's own developer posts, model files on Hugging Face and write-ups from independent reviewers.
Meta's posts cover building with Muse Spark on the Model API, running Muse Glimmer on one GPU and Muse Code leaving beta. Many entries are Muse Glimmer weights and quants, including the official GGUF, Unsloth and bartowski builds, Red Hat AI's FP8 and INT4 versions, and the DFlash drafters.
Reviews and tutorials fill out the rest. eesel covers what Muse can and can't do, Lenny's Newsletter reviews the consumer agent UX, Artificial Analysis publishes its benchmark pages, and HolaClaw tests whether base M3 and M4 Macs can run Glimmer. RuntimeWire runs head-to-head evals of Muse Spark 1.1 against Claude Opus 4.8 and GLM 5.2, and even reverse-engineered the Muse Code binary. KGP Talkie compares Glimmer with Qwen and Nemotron on one RTX 5090. Third-party claims are the authors' own.
Frequently asked
+Where is the official Muse Spark API documentation?
Meta's developer docs are at dev.meta.ai, and the Build with Muse Spark guide on this shelf covers first calls and agentic patterns. The API base URL is https://api.meta.ai/v1.
+Which Muse Glimmer quant should I download?
It depends on your hardware. Meta publishes an official GGUF, Unsloth lists a 17GB 4-bit build, and there are MLX builds for Apple Silicon and FP8 or NVFP4 builds for vLLM. Compare the entries here against your memory budget.
+Is there an independent review of the Muse agent?
Yes. eesel AI's hands-on review covers the Secure VM, security model and pricing tiers, and Lenny's Newsletter reviews the consumer experience.