Skip to main content
Latest Update
2026-11-09 10:04 UTC

Meta unveils Muse Ember with 1M-token context and multimodal IO

Meta unveils Muse Ember with 1M-token context and multimodal IO

According to AI at Meta on X, Meta introduced Muse Ember, a compact model for agent workflows with a 1M-token context window, multimodal understanding, tool use, testing, and targeted code fixes.

Source

Analysis

According to a post from AI at Meta on November 9, 2026, the company introduced Muse Ember as its smallest and most efficient model designed for complex reasoning execution and everyday agent work.

Muse Ember supports carrying workflows across apps, sequential tool calls, running tests and targeted code fixes while handling a 1M-token context window with native understanding of text, images, audio and video for large document sets and varied inputs. This enables developers to reshape workflows by automating multi-step agent tasks and processing extensive multimodal data in one session, with concrete implications for the broader AI ecosystem through more efficient on-device and agent-based applications.

AI at Meta

@AIatMeta

Together with the AI community, we are pushing the boundaries of what’s possible through open science to create a more connected world.