Meta says Muse Glimmer has 30B parameters and is a distilled version of Muse Spark 1.2; Meta plans a $1B fund to invest in US communities near its data centers
Meta Platforms Inc. introduced a new AI model that's light enough to run on a single computer, allowing users to download and customize the technology.
Muse Glimmer extends that arc with a smaller, downloadable derivative of Spark 1.2, while the community fund addresses the local footprint of Meta’s expanding data-center buildout. The two moves show Meta working on both where its models run and how its infrastructure expansion is received.
First-order effects
Developers and users can run and customize Meta’s 30B-parameter Muse Glimmer on a single computer, creating a lower-footprint option derived from Muse Spark 1.2.
US communities near Meta data centers become eligible for a planned $1B investment fund, giving Meta a direct local-investment vehicle alongside its facilities expansion.
Second-order effects
Muse Glimmer gives Meta a self-managed distribution route alongside its aggressively priced Model API, requiring the company to distinguish downloadable customization from hosted model access.
The fund links Meta’s data-center presence to community spending, making local benefits a more visible part of the company’s infrastructure proposition as it finances projects such as Hyperion.
Third-order effects
If Meta maintains this approach, its AI stack will span customizable local models, hosted API access, and capital-intensive centralized infrastructure rather than depend on a single model-delivery channel.
Meta’s combination of off-balance-sheet data-center financing and community investment points toward AI infrastructure projects being evaluated on both financing structure and their local economic commitments.
The trend: AI companies are increasingly pairing multi-channel model distribution with financed data-center expansion and explicit community-investment programs.
Muse Glimmer 30B is a new open source model from Meta. Apache 2.0 license. Capable of local agentic work, code, and image understanding. Fits right on your laptop. It is the strongest model of its size class we've tested. Available now in LM Studio! 🚀🔥 [image]
Build with Meta's new open-weight model on AMD. AMD is enabling day-zero support for Muse Glimmer, the new open-weight model from Meta Superintelligence Labs on AMD Ryzen™ AI Max+ systems and Radeon™ AI PRO R9700 GPUs. With @lmstudio making local AI accessible and Lemonade [image…
LFG: Meta's Muse Glimmer is a remarkably capable model for its size: 30B parameters, local deployment, and the best reported result on 12 of 24 benchmark rows against Gemma4-31B and Qwen3.6-27B. By my count, it beats Gemma on 19 of 24 rows and Qwen on 14. Its strongest area is [i…
Muse Glimmer - the new open-source 30B model from Meta is pre-trained on the outputs of Muse Spark using logit distillation. Also, its performs best for its class size (when compared to Gemma4-31B / Qwen3.6-27B) [image]
@AIatMeta's Muse Glimmer (30B dense, open-weights) launches with SGLang day-0 support. We got ~230 tok/s on a single RTX 5090, with NVFP4 + DFlash on. It also works out of the box on @NVIDIAAIDev RTX Pro 6000, DGX Spark, and MLX for Mac. Speed and reliability have always been [vi…
Meet Muse Glimmer: an open-weight model built for always-on local agents. 30B parameters, Apache 2.0 and tuned for complex multi-step work so it plans, calls tools, hits errors, retries, and sees the task through long-horizon loops. Muse Glimmer is designed to balance [image]
.@andrewrsorkin: Meta “announcing a $1 billion fund aimed at investing in communities close to where the company operates data centers... The Future is for Everyone Fund will be focused, from what I understand, on policemen and firemen and members of these communities.” [video]
Me last month: Hey Mark Zuckerberg - if you want people to feel good about your data centers, write them a check. Zuck today: We're writing $1 billion worth of checks to make people feel good about our data centers. [image]
.@FoxBusiness: @Meta announced a $1 billion Future is for Everyone Fund to invest directly in communities, backing teachers, law enforcement, local infrastructure. Zuckerberg says he was inspired by teachers in Richland Parish, Louisiana who received bonuses thanks to the [video]
So much good stuff in here, particularly around the importance of America leading the global AI race; all of it stemming from having a fundamentally optimistic view of AI and how it will empower people. [image]
Meta seems to be the only company building data centers that knows it needs to show communities there are positives as opposed to scaring the hell out of everyone like Anthropic or being hyper out of touch like Sam Altman.
Good. Communities hosting massive data centers should actually benefit — teachers, law enforcement, and local infrastructure included. More of this from Big Tech.