
Meta has announced Muse Glimmer, a 30-billion-parameter open-source AI model optimised for local agent workflows. The model, released under an Apache 2.0 license, can run on a Mac or PC with a…
Meta has announced Muse Glimmer, a 30-billion-parameter open-source AI model optimised for local agent workflows. The model, released under an Apache 2.0 license, can run on a Mac or PC with a single consumer GPU. It supports local coding, function calling, and LLM-as-a-judge evaluation, and was trained on data covering more than 100 languages, according to gadgets360.com. The Hindu reports the launch comes weeks after Chinese lab Moonshot AI released its Kimi K3 model, which has 2.8 trillion parameters. Meta also plans to open the weights of its Muse Spark 1.2 model as part of a broader push for open-weight AI.
The hype around AI parameter counts and personal superintelligence masks a simpler truth: most users just need models that run reliably on their own devices. Zuckerberg’s vision of a PhD-level tutor for everyone is aspirational but ignores the cost of hardware and data. The real test is whether Muse Glimmer can handle real-world Indian coding tasks on a mid-range laptop without crashing. If it does, open models gain a practical edge. If not, the parameter race will remain a spectator sport.
Sources (2): gadgets360.com, thehindu.com
This story was synthesised by AI from the 2 sources linked above.
Updated: this story now draws on 2 sources.