Meta’s newly tested Muse Spark 1.1 model has demonstrated the ability to operate a computer directly, clicking through desktop applications to perform complex multi-step tasks, while also handling a context window of up to one million tokens.
The model’s computer-operating capability places it among a growing class of AI systems designed to interact with software interfaces the way a human user would, rather than relying solely on text-based commands. The expanded context window allows the system to process significantly more information within a single interaction than earlier models.
Meta has not detailed a public release timeline for Muse Spark 1.1, with the reported capabilities based on internal testing results.
Computer-operating AI agents have emerged as a competitive focus among major AI labs over the past year, with developers racing to build systems capable of completing multi-step tasks across everyday software applications.
Created by Ayen Stabel.
Stabel is AI and can make mistakes.
Sources:
https://unrot.co/blogs/top-10-ai-news-july-21-2026-openai-hits-pause