Meta Releases Muse Glimmer, a 30-Billion-Parameter Open-Weight AI Model

Meta released Muse Glimmer on August 10, 2026, a 30-billion-parameter open-weight AI model built for local deployment on consumer hardware such as laptops and PCs.

The model was developed by Meta Superintelligence Labs and is designed for autonomous agent workflows, including multi-step planning, error recovery during tool execution, and workflow automation. It also features a perception encoder that allows it to process interleaved text and image inputs, enabling it to interpret documents, charts, and screenshots.

Muse Glimmer supports a 128,000-token context window and operates in more than 100 languages.

To run on consumer hardware, Meta compressed the model to approximately 4-bit precision, allowing it to fit within a 24GB to 32GB memory envelope. The model also uses a lightweight companion network to improve generation speeds for long reasoning chains.

Meta trained Muse Glimmer through a distillation process using outputs from its larger model, Muse Spark, followed by supervised fine-tuning and reinforcement learning.

Independent assessments noted a high hallucination rate of about 82% in knowledge-based tasks and said the model trails competitors like Qwen 3.6 27B in coding and operating system-level tasks.

The model is available under the Apache 2.0 license on Hugging Face and supports deployment through llama.cpp, MLX, and ExecuTorch.

Meta CEO Mark Zuckerberg said the release is part of a broader strategy to promote open-source AI. The company also announced plans to release open weights for its more powerful model, Muse Spark 1.2, in the near future.