Meta has released Muse Glimmer, a 30-billion-parameter open-weight AI model designed to run locally on consumer hardware, alongside a lengthy essay from CEO Mark Zuckerberg outlining his vision for "personal superintelligence." The move positions Meta firmly in the open-source camp of the AI debate, contrasting with the proprietary models from OpenAI and Anthropic.
Muse Glimmer is distilled from Meta's larger, closed Muse Spark model and is available under the permissive Apache 2.0 license. It is optimized for agentic tasks like scheduling, file management, and coding, and can operate without an internet connection. Meta says it will also open the weights for its more powerful Muse Spark 1.2 in the coming weeks.
In his essay, Zuckerberg argues that superintelligence should be distributed widely rather than centralized. "Rather than centralizing superintelligence, we should distribute it widely and give every person the ability to direct it," he wrote. He frames this as a democratic counter to the risk of a few companies controlling AI, and positions open models as a way to compete with Chinese labs like DeepSeek and Moonshot.
Reception has been mixed. Some, like Box CEO Aaron Levie, see the move as a "very firm flag in the ground that America will have near-frontier open-source models." Critics, however, question whether the promise of "personal superintelligence" is just branding for Meta's data-harvesting business model, and note that open-weight models often lag behind closed ones in performance.
