GPT-6 Astra IS NERFED!? These Demos Have Me Confused
Did OpenAI nerf GPT-6 Astra, or was something broken? People on X have been posting some pretty wild comparisons, and now Tibo at OpenAI has shared what the team found and fixed. Let's pull up the posts and look at the demos. I haven't noticed a drop myself, but I can see why some of these results have people asking questions. Has Astra felt worse for you? Have the fixes helped? Let me know in the comments. And man, oh man, that's only the start. We've also got Dario Amodei talking about AI helping build better AI, Google DeepMind self-improvement rumors, my Grok 4.7 prediction, GPT-6 Sol demos, Kimi K2.8 Preview, OpenAI's math announcement, and a GPT-Live-1 demo you can actually interrupt. CHAPTERS 0:00 Did Astra get nerfed? 0:26 The comparisons and game demos 1:40 What OpenAI found and fixed 2:27 Dario on recursive self-improvement 3:19 Google's self-improvement rumors 3:47 My Grok 4.7 prediction 4:08 GPT-6 Sol demos 4:22 Kimi K2.8 Preview 4:38 OpenAI's math claim and Buckmaster's response 5:05 Bel and Aeon speculation 5:19 GPT-Live-1's interruption demo SOURCES & DEMOS Astra launch vs. now — noclipepe: https://x.com/noclipepe/status/2098178023790936291 Gun comparison — noclipepe: https://x.com/noclipepe/status/2098323035874828424 Death Star at launch — Lyra: https://x.com/lyraxana/status/2094010424685396403 Death Star nine days later — Lyra: https://x.com/lyraxana/status/2099205310065840199 Rainy-city follow-up — noclipepe: https://x.com/noclipepe/status/2099048926963802455 Game recreation demos — noclipepe: https://x.com/noclipepe/status/2098545145381605431 https://x.com/noclipepe/status/2098870208970846233 Lava-lamp comparison — noclipepe: https://x.com/noclipepe/status/2099153998032871818 Astra fixes and reset — Tibo: https://x.com/thsottiaux/status/2098612714704891959 Dario Amodei — We Must Pace the Frontier: https://darioamodei.com/post/we-must-pace-the-frontier Sam Altman's response: https://x.com/sama/status/2098811563415150910 Elon Musk's response: https://x.com/elonmusk/status/2098789109980332057 Demis Hassabis's proposal: https://demishassabis.substack.com/p/a-framework-for-frontier-ai-and-the-dawning-of-a-new-age Google rumor posts — Lyra and Lentils: https://x.com/lyraxana/status/2098492030787727804 https://x.com/Lentils80/status/2098536329520840986 https://x.com/lyraxana/status/2098536882313285918 Grok 4.7 — Elon Musk: https://x.com/elonmusk/status/2098462085973741960 Sol demo — soliman: https://x.com/letdarky/status/2099235498833113586 Sol model post — Lyra: https://x.com/lyraxana/status/2098101491970031667 Kimi Code documentation: https://www.kimi.com/code/docs/en/kimi-code/models.html OpenAI's math announcement: https://x.com/OpenAI/status/2097374640582668336 https://x.com/OpenAI/status/2097374643518640382 Tristan Buckmaster's statement: https://cims.nyu.edu/~tristanb/statement.pdf Bel and Aeon speculation: https://x.com/Mr_Salio/status/2096233643500831034 https://x.com/thedoomguy_ai/status/2099181359465947567 GPT-Live-1: https://openai.com/index/introducing-gpt-live-1-in-the-api/ #AINews #GPT6 #OpenAI
