Spent an entire weekend burning through 70% of this week's quota—here are my honest thoughts on the GPT-6 Astra model: - First off, the speed is noticeably faster. - That said, the intelligence has improved, but not as much as I expected. Feels more like 5.7 or 5.8 rather than a full 6. - Sometimes it gets lazy. I tried all the tiers—High, Very High, Ultra—and the depth of thought is still a bit shallow. Maybe this is just the GPT-6 vibe: optimized for task completion rather than deep thinking. It’s a totally different feel compared to A Corp’s Fable 5 or Opus 5. Claude, when analyzing stuff, is super meticulous and thorough, trying to cover all angles. Now GPT-6 seems to focus on speed—almost to the point of being sloppy?? - Also, I’ve seen people on Twitter saying GPT-6’s aesthetic sense has improved significantly. Haven’t tested that yet. Conclusion: Sorry, Ultraman, I didn’t see a nuclear explosion, and it didn’t leave me slumped in my chair.