
Ep 236: Alibaba's Qwen4-Max packs 2.4T parameters into a 71B-active MoE with 1M context for…
16.11.2026
0:00
14:29
Models & Agents
Alibaba's Qwen4-Max packs 2.4T parameters into a 71B-active MoE with 1M context for builders today.
What You Need to Know: Alibaba released Qwen4-Max, a proprietary 2.4T-parameter MoE model with 71B active parameters, native vision, and a 1M-token context window priced at $1.5 per million input tokens. Hark shipped Handoff², a computer-use agent that runs twice as fast as its predecessor and completes 82% of 478 internal web tasks. ...
Sources: marktechpost.com · openai.com · securityaffairs.com · news.bloomberglaw.com · arxiv.org · finance.yahoo.com · x.com
Chapters:
0:00 Introduction
0:21 Top Story
0:53 Physicist shares GPT-6 Astra work on Standard Model lattice
1:15 Hark Releases Handoff², a 2x Faster Computer-Use Agent for Hark Pro
2:21 Poetry 'resurrection' from LLMs
2:57 Also Today
12:09 Things to Try
12:43 On the Horizon
13:10 Closing
AI Disclosure: This podcast is curated by Patrick but uses AI-generated voice synthesis for audio production.
🎬 Watch on YouTube (public from 8:15 AM ET): https://www.youtube.com/watch?v=95HzTmKFb9w
If this episode was useful, a rating or a short review on Apple Podcasts or Spotify is how the next listener finds the show — and following it in your app means the next episode is there when you are.
📝 Full show notes, transcript & sources: read the episode page
🎧 Hear this show with the others you follow in one private morning episode, read to you by name. Try My Nerra free for 7 days, then $4.99 a month.
📬 The free Nerra Daily email, every morning at 7 AM Eastern: sign up
🌐 Part of the Nerra Network — explore every show at nerranetwork.com.
Więcej odcinków z kanału "Models & Agents"



Nie przegap odcinka z kanału “Models & Agents”! Subskrybuj bezpłatnie w aplikacji GetPodcast.








