1093 musecases built with Meta Muse

@Telep-IO

@Telep-IO

Omusey is an Omarchy plugin that puts Muse chat in the Linux desktop bar as a dropdown that can pop out into its own window.

@NeerajMohanty

@NeerajMohanty

Open-source, stateless MCP connector that publishes explicitly approved text posts from Muse to LinkedIn, with two tools and no database or scheduler.

AaryanK

AaryanK

Solo-built GGUF line of Muse Glimmer 30B with custom calibration, per-tensor allocations and an eval harness behind every reported number.

thehype.

@thehypedotnews

gemini 3.7 flash vs deepseek v4 pro 0813 vs muse spark 1.2 – on voxel city dioramas three models each built three crossy road-style 3d scenes – a construction site, a nyc intersection, a river with a drawbridge – as single self-contained html files the setup: @nousresearch's

Picked up China version of the Mi50 (Radeon VII) 16GB VRAM GPU for about $135. Ran it using llama.cpp Ubuntu Vulkan prebuilt binary build: b29c606e2 (10964). Used a Power Limit or 220/190 watts on the GPUs. Dual Radeon 32GB Vram and 64GB DDR4 System Dual Radeon RX 7900 GRE and Radeon VII 32gb VRAM GGUF Models: • Qwen3.6-35B-A3B-UD-Q5_K_XL.gguf • Accio-Lab_occamy-1.0-Q5_K_S.gguf • Laguna-XS-2.1-APEX-I-Balanced.gguf • NVIDIA-Nemotron-3.5-Lightning-30B-A3B-Q5_K_M.gguf • Gemma-4-31B-it-Q6_K.gguf • Ateron_Gemma-4-MoonGem-31B-Q5_K_M.gguf • Qwen3-Coder-30B-A3B-Instruct-UD-Q6_K_XL.gguf • Qwen3-VL-30B-A3B-Thinking-UD-Q6_K_XL.gguf • Qwen3-Coder-30B-A3B-Instruct-UD-Q5_K_XL.gguf • North-Mini-Code-1.0-MXFP4_MOE.gguf • GLM-4.7-Flash-Uncen-Hrt-NEO-CODE-MAX-imat-D_AU-Q6_K.gguf • Muse-Glimmer-30B-UD-Q6_K_XL.gguf • Huihui-Qwen3.8-27B-abliterated-UD-Q6_K_XL.gguf • Qwen3.8-27B-Q6_K.gguf • Qwen3.8-27B-OBLITERATED-Q5_K_M.gguf • Medgemma-27b-it-UD-Q6_K_XL.gguf • Gemma-4-26B-A4B-it-UD-Q6_K_XL.gguf • Gemma4-26B-A4B-QAT-Uncensored-HauhauCS-Balanced-Q4_K_M.gguf • GPT-OSS-20b-abliterated.i1-Q6_K.gguf Sorted by params then size model size params pp512 tg128 qwen35moe 35B.A3B Q5_K - Medium 24.76 GiB 3

Muse Glimmer 30B feels significantly more precise and reliable, it almost never drops the ball or breaks rules. However, its designs lack creative depth and richness. Qwen3.6 35B, on the other hand, is prone to more occasional blunders/hallucinations, but its creative output is superior. It generates far richer, more complex voxel worlds and offers higher design quality. LLama.ccp Build Provenance: • Base: llama.cpp upstream (merge 4445f8d, build 661) • CUDA Toolkit 13.1 + MSVC 19.44 + sm_120a-real (native Blackwell PTX) • Flags: GGML_CUDA=ON, GGML_CUDA_FA=ON, GGML_CUDA_FA_ALL_QUANTS=ON, GGML_CUDA_GRAPHS=ON, GGML_NATIVE=OFF • License: MIT (upstream llama.cpp) Do you think Qwen3.6 is still the undisputed king here?

Connor Loi

@connortbot

You can now run Muse Code (and Muse Spark 1.3) in cloud agents on @tryreplicas! Bring your own Muse subscription - they're surprisingly generous and powerful.

dev.meta.ai

dev.meta.ai

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.

Amjad Masad

@amasad

Muse can now make apps on Replit

@axlecoffee

@axlecoffee

A BYOK VS Code extension that registers the InferHub catalog, including Muse Spark with its 1,048,576-token context window, as a model provider in the Copilot Chat picker without Copilot Pro.

@albertodepaola

@albertodepaola

Out-of-tree Hermes Agent model-provider plugin for the Meta Model API and the Muse Spark family, including the low-cost muse-spark-1.2-contributor tier.

@CogniTechSystems

@CogniTechSystems

This repo documents running Muse Glimmer 30B locally on an M4 Max MacBook via llama.cpp, benchmarking it with and without speculative decoding, and wiring it into Claude Code through LiteLLM for fully offline coding.

@EclipseAditya

@EclipseAditya

Pi coding agent extension that adds Muse Spark 1.2, 1.2-contributor and 1.1 via the Meta Model API, a maintained fork that fixes a false auth warning on pi v0.84+.

WEBdoze pairs Muse Spark 1.3 and Gemini 3.8 Flash with the Impeccable design skill and browser testing to turn plain AI-generated pages into polished frontends.

darkc0de

darkc0de

Decensored Muse Glimmer 30B made with Heretic v1.4.0, shipped with a reproduce directory.

I'm guessing that many people have been waiting for this comparison. For clarity, both models are running at full FP16 KV-cache. Due to VRAM limitations, Muse Glimmer is running full 262,144 context, whilst Qwen3.6 27B can only run at 147,500 context - full GPU offload in both cases. Both models have been coding on an enterprise-grade web application. Detailed report of each model (warning - includes AI generated content): Diagnostic quality - comparable. Both have shown genuinely good root-cause work when they apply themselves. Qwen found coding issue and worked to fix things cleanly. Muse Glimmer correctly traced bugs and even caught something that a Frontier model missed after more than 10 rounds of review. Neither one is weak at diagnosis. Implementation reliability - Qwen ahead. Qwen did introduce real regressions into the coding along the way (eg. severe zone-scope refactor regression, and case-sensitivity regression) but each one eventually got fixed properly once caught, usually within one or two corrective rounds. Muse Glimmer did land fixes that were clean and verified true to spec. However, when working in a complex environment exceeding 200k context, Muse Glimmer fa

🤙🏼🇲🇽Hela🇺🇸🤙🏼

@Mexi_Coug

👀👀 @Muse keeps impressing me!! Doing the personal stuff was great but today I decided to test it and asked it to help me with some work stuff. It created its own CRM, categorized leads into different tiers and also created a sub-agent that will help with outreach. It did all h

DHH

@dhh

I had Muse Code convert a TanStack version of our new site to Astro. It did a flawless job. And all I managed to spend was 20 cents!! I think maybe this token stash from @finkd and @AIatMeta is going to last longer than the sun has gas left to shine 🌅

Boz

@boztank

Very excited about the launch of @Muse today. I’ve been using it internally for months and I am hard pressed to think of any product that I’ve come to rely on more in such a short period of time. I have it linked to my email, calendar, and credit cards. I use it to help me plan

Tom Collins

@thetomcollins

$META just went from 3.5% to 45.4% token share on OpenCode in just over two weeks Muse Spark 1.3 being good + free is enough to become the default for most users Default gets you usage → usage gets you data → data makes the next model better Anthropic and OpenAI can’t afford

deeplearning.ai

deeplearning.ai

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.

Hi all, Profile v2.2 is out. It's an open-source optimizer for inference servers. It computes your GPU's roofline ceiling, measures your live server against it, names the bottleneck, gives the flag. You apply. It re-measures. Every fix answers to a number. vLLM only today. More engines next. This release: core rule engine rewritten. Eight rules on a priority DAG with mutual exclusivity. Five alarms fire, four echoes are silenced, one true cause survives. Deterministic. AMD cards are now supported too. Tuning today is chaos: try a flag, wait, squint at a dashboard, repeat for days. Profile turns it into deterministic engineering: measure, fix, verify. Results in a few iterations. Mine took 4, ~30 minutes. My setup: RTX 5090, muse-glimmer 30B, SWE-Bench agents, no DFlash spec decoding. • 81 → 421 tok/s at 25k ctx • $3.41 → $0.65 per 1M output tok • TTFT 224ms (p95 500ms), TPOT 23ms at end of run • 4.72 → 1.08 J/tok https://preview.redd.it/4vazyxkcq6kh1.png?width=2248&format=png&auto=webp&s=77923a489b6f725240d23a7953150b5779260734 One iteration regressed hard: KV thrashing, TTFT 32.8s. Profile labeled it worse. Next fix recovered it. Regressions stay in the record. Watc

mihawkxxx

@mihawkxxxxx

Incredible Minecraft from muse spark 1.3 cost 10 cents

A harder rematch between Muse Glimmer and Qwen 3.8 27B (bumped to high reasoning) on the TalkWithMe project.

@ferrreo

@ferrreo

Custom firmware and a Go server that turn an M5Stack StackChan into Tarquin, a voice-driven robot butler with wake word, face tracking and on-device speech, using Muse Spark 1.3 via the Meta Model API as its brain.

チャエン | デジライズ CEO《重要AIニュースを毎日最速で発信⚡️》

@masahirochaen

【速報】Metaが30BのオープンモデルMuse Glimmerを公開。18GBのRAMで動く。 久々のMetaからの本格オープンモデル。 小型なので低スペックのPCでも動かせるAIモデル。 ・Apache 2.0で商用利用可、重みはHugging Faceで公開 ・画像も読めるvisionモデル、100言語超に対応 ・MCP Atlas

Arena.ai

@arena

Muse Spark 1.3 (xHigh) just landed @AIatMeta back in the top 10 models for Code Arena: WebDev! This release is ~#10 in Code Arena: WebDev with 1623 pts (AutoEval). It’s performance is par with Claude Fable 5 at #8 (1628 pts) and GPT-5.6 Sol at #11 (1616 pts). Muse Spark 1.3

Cline

@cline

The successor to Llama is here, and Meta is revitalizing focus on open weights with their new Muse Glimmer - a leading 30B param model designed for always-on local agent use, small enough to run on a Mac or PC with a single GPU. Available in Cline using Ollama now!

Deep Mehta

@mehtadeep

.@Muse is incredible! Ran 146 flight price searches (cash and award flights w/ points) from SFO to 11 cities in Spain and departing from 10 Italy cities to SFO across 7 dates (I had a flexible travel schedule) to give me an optimized lowest round-trip flight cost!-

@MiaAI-Lab

@MiaAI-Lab

A gallery of 100 self-contained generative-art, physics-toy and typography pages generated with Muse Spark 1.3, each shown next to its original prompt.

Muse

@Muse

Get more done with Muse with two new connectors: ✅ @meetgranola: your meeting notes ✅ @NotionHQ: your docs …and more coming soon! Ask Muse or connect in settings.

Alexandr Wang

@alexandr_wang

1/ releasing muse image today — the first image generation model from MSL. it's agentic: pairs with muse spark to reason through your prompt, search the web, and plan before it generates. people get what they meant on the first try. live now in the Meta AI app.

+3

Paolo Rosson

@redp314

Got Meta's new Muse Glimmer 30B running on my MacBook (M3 Max, 96GG) and tested the serving options available so far. Fastest right now: Ollama's MLX engine (DFlash included) at ~29 tok/s. Tuned llama.cpp: ~21. Raw mlx-vlm: ~10, not optimized yet. Numbers below if you're

Feifei Qiu

@feifei_qiu

@Muse saved $720 for me tonight in an hour Google AI Pro : I have been paying $19.99/mo for a long time, while I have Verizon perk not being used (free 6 months, then $10) → ~$180 saved in the first year McAfee auto-renew for an old laptop: canceled → $200/yr avoided

@dttdrv

@dttdrv

An unofficial native macOS client for the Muse Code CLI in Meta AI's design language, sharing sessions with the terminal so work can resume in either place.

@anuragdhar

@anuragdhar

A VS Code extension that runs the Muse Code CLI in a sidebar and also adds Muse Spark 1.2 models to VS Code Chat Agent Mode, with inline completions and live token and cost tracking.

opentools.ai

opentools.ai

OpenTools maintains a source-backed directory of about 163 Muse connector setup guides, separating Meta-documented, partner-documented and community skills (e.g. HubSpot, Attio, Ashby, Plaid, Slack), with status labels such as draft or maintainer-tested.

@youdotcom-oss

@youdotcom-oss

You.com's harness for evaluating Muse Glimmer 30B on DeepSearchQA with You.com MCP tools inside pi sessions. A custom RLM v5 extension reached F1 0.8054 on 50 tasks (0.6910 over 900x3) versus 0.50 for plain skill injection.

kwindla

@kwindla

Exciting to see Meta releasing new open weights this week. Meta trained the new 30B dense model Muse Glimmer with "agentic" use cases in mind. This generally means task-oriented, multi-turn, and heavy use of tool calling. I ran Muse Glimmer in the GGUF quant through a bunch of

@bentedesco

@bentedesco

A shared Socket Mode transport between Slack and Muse bots: one listener daemon appends Slack events to a JSONL log, a CLI posts replies back, and each bot runs its own consumer with no public HTTP endpoint.