Meta returns to open models with Zuckerberg's plan to out-copy China and sell compute by auction
Meta released Muse Glimmer, a 30B open agent model from its new Superintelligence Labs that runs on consumer hardware. Mark Zuckerberg also defended distilling models to counter Chinese competition.
- Meta released Muse Glimmer, a 30B open agent model requiring under 20GB RAM.
- Zuckerberg defended distilling rival models to ensure US leadership over China.
- The release comes from the new Meta Superintelligence Labs division.
- An open-weight version of Muse Spark 1.2 is expected soon.
Meta has introduced Muse Glimmer, the first open model from its newly formed Superintelligence Labs. This 30 billion parameter agent model is designed to run on consumer hardware, requiring less than 20 GB of memory once weights are compressed.
In a strategic essay accompanying the release, Mark Zuckerberg argued for fewer restrictions on US AI labs. He explicitly defended the practice of distilling models from other companies, framing it as necessary to maintain American leadership against China.
The move signals a return to aggressive open-sourcing by Meta, contrasting with the closed approaches of competitors like OpenAI and Anthropic. An open-weight version of Muse Spark 1.2 is reportedly expected to follow soon.
Provides access to a powerful 30B agent model capable of running locally on consumer-grade hardware.
Signals Meta's commitment to open weights, potentially lowering costs and increasing competition against closed-source vendors.
Highlights Meta's dual strategy of open-sourcing models while building compute infrastructure for auction.
Intensifies the debate between open and closed AI development models.
- Distillation
- The process of training a smaller, efficient model to mimic the behavior of a larger, more complex model.
- Weights
- The internal parameters of a neural network that determine how it processes input data to generate output.
Open SourceI counted every Bun bug before and after the Rust rewrite
Open SourceMeta is back with Muse Glimmer: local, agentic, multimodal, and open source
Open SourceCodeberg banning AI code is not a cult. It's a boundary.
Open SourceDeploying Dify – Open-Source LLM Application Development Platform
Cloudflare open-sources vibe-coding platform for people who aren't coders
NCCU opens nation’s first HBCU artificial intelligence insti... - QCity Metro
North Carolina Central University has opened the first artificial intelligence institute at an HBCU, aiming to advance AI education and research for underrepresented groups.
'AI is for everyone:' NC Central opens artificial intelligence institute - WUNC News
North Carolina Central University has opened a new artificial intelligence institute aimed at making AI education and research more accessible to all students.
Artificial Intelligence: Moving up the Investment Value Chain - ai-cio.com
Institutional investors are moving beyond pilot projects to full-scale AI allocations, reshaping the investment landscape.
AI ResearchAI professors are negotiating the new realities of academic research
Top AI professors gathered in Mountain View to discuss adapting academic research to the fast-changing realities of AI development.
With a feel for physics, AI models simulate a wider range of real-world scenarios - news.mit.edu
MIT researchers have created AI models that incorporate physical laws, allowing them to simulate a broader array of real‑world scenarios.
Artificial Intelligence in Dermoscopy: Why Expert Oversight Still Matters - Medscape
A Medscape analysis argues that AI tools for dermoscopy still require expert oversight to avoid misdiagnosis in skin cancer detection.