Meta Superintelligence Labs has introduced Muse Glimmer, a 30-billion-parameter artificial intelligence model designed to run AI agents directly on consumer computers, expanding access to advanced AI without continuous reliance on cloud infrastructure. The company is releasing the model’s weights under the permissive Apache 2.0 licence, allowing developers to download, modify and build applications for local AI use cases. Meta said Muse Glimmer was developed for “always-on” local agent workflows, with applications spanning AI assistants, function calling, coding, document and file management, and large language model evaluation. The model operates on a Mac or PC with a single consumer GPU, enabling deployment on personal devices rather than remote cloud servers. Meta stated that running models locally enables AI use anywhere, anytime, with or without an internet connection. Muse Glimmer is available through Hugging Face alongside developer documentation. Unlike conventional AI models, it handles extended sequences of tasks and interacts with external tools, combining long-horizon execution, tool calling, multimodal understanding, long-context memory and instruction following. It can make function calls, reason through multi-step tasks, recover from failed tool calls, and work with text and images, including screenshots, charts and documents. Trained on data covering more than 100 languages, it was evaluated on benchmarks including DeepSearch QA, MCP-Atlas, τ-Bench and SWE-Bench, recording strong results for its size category compared with models such as Gemma4-31B and