GPT 6.1 "Near Astra" and Usage limit in Plus and Pro
Questions for Pro/Plus subscribers who've tried the new usage limits, Pro subscribers who've used Astra, and anyone who's already tested GPT-6.1 I'd love your take
I haven't used Codex since GPT-5.6. I'd been waiting for the $200 subscription to come back so I could use Astra to finish projects where GPT-5.6 kept getting stuck (on Plus I also ran out of usage really fast). While I was waiting, I ended up in the hospital, and during that time a whirlwind of models dropped, including GPT-6 Sol and GPT-6.1.
I was thinking about buying the $200 plan, but the usage limit cut announced at DevDay doesn't convince me. Then again, given Sol's efficiency and intelligence improvements, I'm not sure I'll even need it. So I have two questions:
- Usage: According to AA (v4.3.2), the total cost to run the index dropped 68.8% from GPT-5.6 to GPT-6.1, both at max effort. Does that actually translate into ~3.2x more usage in Codex?
- Intelligence: For those who have used Astra for a while on really complex tasks (research, ultra mode, computer use, 3D modeling) and have also given 6.1 Sol a real shot: is it actually as smart as Astra? On AA v4.3.2, 6 Astra ExtraHigh and 6.1 Sol Max are tied. Can 6.1 Max really match Astra ExtraHigh in practice?
Thanks in advance!