Where Muse Spark aimed to be a lightweight model that could perform at low latency, the Wonder models are larger and said to be more capable, with higher benchmark scores and better results on complex tasks like coding.
💡 Why it matters

Editorial read: "Meta releases its firs..." is a research progress and benchmark changes update; source signal: TechCrunch AI as a news portal, AI 80, still inside the 24-hour attention window; value: following technical direction and evaluation shifts.