Apple's HomeView at $299, and a benchmark of Meta's Muse Wonder

Hardware and the home: Apple's HomeView arrives

Stratechery: "Apple HomeView, The New Apple TV, The App Story"

1. The free line asks a market question. The summary reads: "The Apple Homeview is official, and the question is how big is the addressable market for expensive devices based on subpar AI." The update lands the day after Apple's launch. HomeView is a $299 screen with a 7-inch display, and it recognizes up to six people by face. 13 Pre-orders opened October 13 and the device goes on sale November 13. Apple lists the European Union as outside the launch, because Siri AI has no iOS release there yet. 3
2. Two of the three markers match hardware Apple announced. The markers are Apple HomeView, The New Apple TV, and The App Story. 2 The second is the new Apple TV 4K, announced the same day at the same $299, with Siri AI, three far-field microphones and a home hub role. 4 The third has no matching press release. Thompson's last free essay on the subject argued that agents are the ultimate aggregators and that providing them is tech's biggest prize. 5
3. Thompson's published view of Apple's home strategy already leaned this way. His October 7 update opened: "Apple is taking a smarter approach to the home than I expected, leaning into integration (with partners)." 6 Apple's own release bears the partnership out. HomeView works with the Apple Home Collection, the cameras, thermostats and lock Apple built with LG Electronics and Schneider Electric. 3

AI models: a builder's bench for Meta's Muse Wonder

Lenny's Newsletter / How I AI: "Meta's new 'Muse Wonder' model: good for AI agents, bad for design"

1. How the test ran. Vo ran Meta's Muse Wonder and Muse Wonder Full headlessly through Claude Code against nearly 20 tasks. The set covers prototyping, agentic workflows, coding, writing and creative work. She scored the results with an LLM-as-a-judge pass plus her own voice notes. 7 Meta released both models on October 12, claiming state-of-the-art results on personal and professional agentic tasks. Both are built on what Meta calls Active Memory and Active Focus. 8
2. The verdict splits by task. The title states the split: good for AI agents, bad for design. 7 The free page lists the comparisons behind it. Muse ranked first on three agentic tasks. The episode also sets it against Sonnet on email writing and voice-assistant work. On SVG illustration and short-form video, the comparison is with Opus 5.5, Astra and Fable. 7 It also reports where Muse falls short, and that the model is fast on one-shot work and slow on long agent tasks. 7
3. The answer a reader is deciding on. The page promises a verdict on whether Vo will put Muse into her own agent workflows. 7 Both models are available now through Meta's Model API and Muse Code, and Meta says Muse Wonder will soon power Muse, its consumer agent. 8

Where to start