AI Circle Daily
https://aicircle.news/archive/2026-10-13?lang=en
Scan to read this page
Meta dropped Muse Wonder, its first genuinely frontier-scale model — bigger and more capable than the lightweight Muse Spark, with higher benchmarks and better results on complex work like coding.
Take: Meta's been playing catch-up long enough; Spark was a placeholder. Wonder is the real 'we're still in this' card, and it drags the whole model arms race up another notch.
Microsoft unveiled the next generation of its custom AI chip and says it plans to deploy it at scale, explicitly eyeing cost savings.
Take: Inference cost is where the real money bleeds, so this move was inevitable. It protects Microsoft's own margins — but fully ditching Nvidia? Not even close yet; watch the fab capacity and software ecosystem.