Next Token Podcast Ep. 2: Is GPT-6 Astra AGI? DeepSeek Flash, Harness and iPhone Duo
op7418 · x · 2026-09-14
The second episode of the "Next Token" podcast (103 minutes, recorded at Shanghai's AGI Bar) is out, co-hosted by 向阳乔木, 歸藏, 橘子 and 杨攀. Key topics:
- GPT-6 Astra: starting from its 3D modeling experience, the hosts debate whether an agent that can operate a computer and complete real tasks qualifies as AGI, with differing verdicts.
- DeepSeek V4.1 Flash and Harness: once speed, cost and capability are good enough, the bottleneck shifts to testing, delivery and human judgment; discussion of cloud agents, resident computers, Agent APIs, and whether small teams should build their own harness or focus on data, interfaces and specific use cases.
- iPhone Duo: beyond the foldable form factor, how screen changes affect app interaction, interface standards and long-term consistency.
- Workflows: the hosts share how they use AI to produce the podcast, personal context memory, voice input friction, controllable music/image editing, and the perennial "are models getting dumber" question.
The episode includes a full timeline and runs about 103 minutes.
Related event: Next Token Podcast Episode 2 Discusses GPT-6 Astra and iPhone Duo(3 posts)→
More from AGI Musings
- Bryan Johnson: I don't trust anyone's AI risk assessment, so I'm betting on optimism — QuanquanGu · 2026-09-14
- What should AI labs do if AGI risk were real? A challenge to skeptics — AndyMasley · 2026-09-14
- What Is Science For? When AI Answers Math Problems, Minds Stop Changing — EdwardSun0909 · 2026-09-14
- Tech billionaires building doomsday bunkers: worrying signal or just prudent spending? — AIandDesign · 2026-09-14
- Karpathy: RL is terrible, but we're building ghosts, not animals — RileyRalmuto · 2026-09-14
- Scott Alexander lays out his full theory of morality and charity amid drowning child debate — john__allard · 2026-09-14