Editorial illustration for LWiAI Podcast #252: GPT 5.6, Grok 4.5, and AI 2040 Discussed
GPT 5.6 and Grok 4.5: AI 2040 Roadmap Discussed
Andrey Kurenkov and Jeremie Harris are back after a stretch of dead air on the Last Week in AI podcast, and episode 252, recorded November 7, 2026, has some catching up to do. Kurenkov opens with an apology for the silent pause, blames an overloaded work schedule, and says paid subscriptions are on hold until posting gets consistent again. Then it's straight into a stacked week of model releases.
OpenAI pushed out GPT-5.6, including variants named Sol and Luna, and renamed its desktop coding agent ChatGPT Work, all while facing disputed claims that US government officials had a hand in green-lighting or delaying the launch. SpaceX AI's Grok 4.5 landed as a cheap, Opus-class coding model with barely any safety documentation attached. Meta countered with Muse Spark 1.1, undercutting on price while posting big coding and cybersecurity benchmark gains, backed by a much longer safety writeup than its rivals bothered with.
Meta also previewed Muse Video and shipped Muse Image, only to pull back fast once users started generating images of public Instagram accounts without consent.
OpenAI publicly rolled out GPT-5.6 (including Sol and Luna) and rebranded its desktop agentic coding product as ChatGPT Work, amid disputed claims about whether the US government effectively green-lit and delayed the release and concerns about inconsistent, ad hoc frontier-model oversight and jailbreakability.
Why this matters
For a podcast that's been the reliable Monday-morning download on AI news, a several-month silent pause is itself a data point worth sitting with. Andrey Kurenkov's note about work overwhelming the schedule, and paid subscriptions getting frozen until posting stabilizes, tells us something about the pace of this industry: even the people paid to summarize it weekly can't keep up. That's worth remembering next time a lab claims real-time transparency about its own releases.
On substance, GPT-5.6's rollout alongside Grok 4.5 and Nemotron-Labs-Diffusion chatter means the update treadmill hasn't slowed just because the newsletter did. If you're building on these models, the lesson isn't in any single feature comparison, it's in the gap between release cadence and anyone's ability to actually track it. We'd take "AI 2040" discussions with a grain of salt from hosts admitting they're behind on last month's news, let alone next decade's.
Catch-up episodes covering old ground are fine, but readers should watch whether Last Week in AI can actually return to weekly cadence, or whether this becomes the new normal for AI commentary generally.
Common Questions Answered
What new model variants did OpenAI release alongside GPT-5.6?
OpenAI released GPT-5.6 along with two variants named Sol and Luna. These releases were part of a significant week of model launches covered in the Last Week in AI podcast episode 252.
What product did OpenAI rebrand as ChatGPT Work?
OpenAI rebranded its desktop agentic coding product as ChatGPT Work. This rebranding was announced alongside the GPT-5.6 release in November 2026.
What concerns were raised about the GPT-5.6 release?
The release faced disputed claims about whether the US government effectively green-lit and delayed the release, along with concerns about inconsistent, ad hoc frontier-model oversight and jailbreakability. These issues highlight ongoing debates about AI model regulation and security.
Why did the Last Week in AI podcast go on hiatus?
Host Andrey Kurenkov cited an overloaded work schedule as the reason for the several-month pause in podcast episodes. He also noted that paid subscriptions were being put on hold until posting could become consistent again, reflecting the difficulty of keeping up with the rapid pace of AI industry developments.
What does the podcast's hiatus reveal about the AI industry pace?
The hiatus demonstrates that even professionals dedicated to summarizing AI news weekly struggle to keep up with the industry's rapid development cycle. This suggests that claims about real-time transparency from AI labs should be viewed with skepticism given the overwhelming volume of releases and developments.
Further Reading
- Last Week in AI #252 - GPT 5.6, Grok 4.5, Nemotron-Labs ... - Last Week in AI (YouTube)
- AI News Went Nuclear: GPT 5.6, GPT Live, Grok 4.5, Muse ... - AI Explained (YouTube)
- Three Models Drop: Grok 4.5, Meta Muse & GPT-5.6 Sol ... - YouTube
- The AI Show Episode 225: GPT-5.6, ChatGPT Work, Enterprise Agents, AI 2040 & Apple Sues OpenAI - SmarterX AI Show
- The Big Model Release Week: GPT 5.6 Is the New Frontier ... - YouTube