Mistral Large 4, day two: 1.05T parameters, 52B active, open weights promised by end of October
The official docs are live: Mistral Large 4 is a granular mixture-of-experts model — 1.05 trillion total parameters, 52 billion active, a 1.6B vision encoder, 1M context, API pricing at $0.68/M input and $2.09/M output on launch sale. Public preview opened October 6; open weights are promised by end of October. The internet has already nicknamed it 'Le Chonk.'
ChatGPT is getting a lot more visual, with a new interface
OpenAI is relaunching ChatGPT around a much more visual interface — generation as canvas, not chat log. The UI wars are officially a front.
Tony Fadell on why the first wave of AI gadgets failed — and what comes next
The man behind the iPod weighs in: Tony Fadell explains why the first wave of AI gadgets flopped and what the next wave has to get right. Hardware's hard — even when the model is good.
Google experiments with an AI-powered gaming platform
Google is prototyping an AI-powered gaming platform — games that talk back, worlds that generate. The search giant keeps circling play.
Meta’s Muse launches on iPad, a month after its mobile debut
Meta's Muse agent spreads to the iPad barely a month after mobile — distribution velocity is the new agent battleground.
ChatGPT for Teens keeps teens talking, even during mental health crises
A study of OpenAI's teen-tier ChatGPT finds it keeps adolescents talking — including through mental-health crises. The finding raises as many safety questions as it answers; the research is early.