Sakhanda Wire
NVDA $152.18 -2.54% MSFT $505.66 -2.38% GOOGL $279.68 -1.43% META $617.65 -1.40% AMZN $256.02 -0.67%
← Back to the news

Meta releases open-source Muse Glimmer multimodal model

Meta has released Muse Glimmer, a 30B-parameter multimodal model designed for local, agentic applications. The model is distilled from Muse and distributed under the Apache 2.0 license, positioning it for privacy-sensitive uses such as coding, document analysis and personal assistants. It is available on the Hugging Face Hub.

Muse Glimmer combines a 2B-parameter vision encoder with a 28B-parameter text decoder. It can process images and videos, make multimodal tool calls and perform open-ended object detection. Its video processor targets two frames per second and supports clips of up to 96 sampled frames.

Meta said the model received day-one support in Transformers, llama.cpp, vLLM, Inference Endpoints and other libraries. An optional DFlash speculative-decoding drafter can increase generation speed, with additional memory use. In the published comparisons, Muse Glimmer led several agentic, coding and reasoning tests, while other models scored higher on some benchmarks, including selected multimodal and safety measures.

Written by Sakhanda Wire AI from reporting by Hugging Face. Facts, quotes and full context are in the original article below.

Originally published by Hugging Face on

Read the original on Hugging Face ↗

The reporting and the image belong to Hugging Face and are credited and linked above; the text on this page is Sakhanda Wire's own summary.

← Back to the news

More stories

All the latest news