News Summary
- Meta released Muse Glimmer on August 10, 2026, a 30-billion-parameter open-weight multimodal AI model licensed under Apache 2.0.
- The company also announced it will open the weights of Muse Spark 1.2, its most advanced AI model.
- Mark Zuckerberg published a 6,500-word essay titled “The Future is for Everyone” making the case for open-source AI.
- Zuckerberg criticized closed AI rivals and urged Washington not to restrict foreign open-weight models.
- Meta announced a $1 billion fund for communities hosting its data centers and committed up to $145 billion in capital expenditures for 2026.
- The release follows the August 5 launch of Muse Code, a terminal-based coding agent powered by Muse Spark 1.2.
Meta released Muse Glimmer on August 10, a 30-billion-parameter open-weight artificial intelligence model designed to run on a personal computer, alongside a pledge to open the weights of its most capable model, Muse Spark 1.2. Chief Executive Mark Zuckerberg published a 6,500-word essay arguing that American open-source models must become the best in the world and criticizing rivals that keep their systems closed.
Muse Glimmer is a multimodal model that combines a large language model with a dedicated perception encoder, supporting both natural language and visual inputs. It is distributed under the Apache License 2.0 and can be used with inference frameworks including Hugging Face Transformers, llama.cpp, Ollama, and vLLM. Meta designed it for local agentic and coding workloads.
The release marks a return to open-weight releases for Meta after a period in which the company mixed open and proprietary offerings. Meta initially built its AI reputation on the open-weight Llama family, launched in February 2023, but more recently focused on proprietary models through its Meta Superintelligence Labs, the unit led by former Scale AI Chief Executive Alexandr Wang.
Zuckerberg’s essay, titled “The Future is for Everyone,” argues that powerful AI should not be controlled by a handful of companies, institutions, or governments. He warned that concentrating control of advanced AI among a select few poses the technology’s biggest risk.
“I do not believe restricting access to foreign open source models is an effective solution,” Zuckerberg wrote. “Our goal should be for American open source models to be the best globally. This requires removing the hurdles that make it harder for American open source models to compete.”
The statement arrives as Chinese startups have taken a leading position in open-weight AI development. Moonshot AI’s Kimi K3, released in July, ranks among the most capable open-weight systems available. Alibaba and DeepSeek have also released competitive open models. Unlike their Chinese counterparts, the top offerings from U.S. firms OpenAI, Anthropic, and Google are not publicly available as open-weight systems.
Zuckerberg’s position puts Meta at odds with some voices in Washington who have called for restrictions on Chinese open-weight models. The Trump administration has accused Moonshot of improperly distilling from Anthropic’s Fable 5, and Treasury Secretary Scott Bessent has suggested the United States could sanction companies found to have used distillation to extract intellectual property.
The essay also outlines Zuckerberg’s vision of a future in which everyone has access to “personal superintelligence” capable of providing Ph.D.-level tutoring, helping create new businesses, and offering personalized lifestyle guidance. He wrote that his eight-year-old daughter can already code her ideas and quickly produce videos using AI tools.
“Everyone will soon have invention superpowers,” Zuckerberg wrote.
Not everyone shares that optimism. Anthony Aguirre, president of the AI safety nonprofit Future of Life Institute, noted that Meta’s own models have been involved in incidents where AI systems hacked other companies on at least three separate occasions. “What on Earth makes Meta think they,or anyone,could control something exponentially smarter, faster, and more capable?” Aguirre said in a statement.
Alongside the model release, Meta announced several infrastructure and community initiatives. The company is launching a $1 billion Future Is For Everyone Fund directed at U.S. communities that host Meta data center facilities. Meta has committed up to $145 billion in capital expenditures for 2026, a figure that reflects the massive scale of its AI infrastructure buildout.
Zuckerberg also addressed the public relations challenges surrounding data center construction. He pledged that Meta would build energy-generating infrastructure “wherever we invest” that could return low-cost energy to host communities, and committed to restoring more water than the company uses in the watersheds where it operates by 2030. Meta said it would provide free training and guaranteed high-paying jobs for skilled tradespeople in data center regions.
The Muse Glimmer release followed closely on the August 5 launch of Muse Code, a terminal-based coding agent built on Muse Spark 1.2. Muse Code can write and debug code, complete longer software development tasks, and run multiple sub-agents concurrently. Muse Spark 1.2 itself is available through the Meta Model API and OpenRouter, with API pricing set at roughly 25 percent of what Anthropic and OpenAI charge for comparable models, according to Zuckerberg.
Meta’s renewed open-source push comes as the company tries to position the Muse family as the foundation for a more capable Meta AI assistant across Facebook, Instagram, WhatsApp, and its Ray-Ban Meta AI glasses. Whether the strategy will close the gap with closed rivals remains an open question. Muse Spark 1.2 currently ranks 47th on the public BenchLM leaderboard with a score of 61 out of 100, placing it behind the frontier models from OpenAI and Anthropic.
For now, Meta is betting that openness, scale, and price will be enough to keep the company competitive in an AI landscape increasingly defined by closed systems.