Meta has released Muse Spark 1.1, a multimodal reasoning model designed for agentic tasks, featuring significant improvements in tool use, coding, and computer interaction. The update introduces advanced capabilities such as managing a 1-million-token context window and orchestrating multi-agent systems to optimize latency. Additionally, Meta launched a public preview of the Meta Model API, allowing developers to access this new model.
Background
Meta continues to advance its AI research through its Superintelligence Labs, focusing on creating models capable of autonomous action and complex reasoning. This release follows previous iterations of the Muse series, aiming to bridge the gap between theoretical superintelligence and practical application.
- Source
- hackernews
- Published
- Jul 9, 2026 at 10:10 PM
- Score
- 8.0 / 10