What the community actually wants from OpenAI DevDay:
1. Fix the limits
After Opus 5.5, everyone saw that Claude, historically known for its strict limits, is now actually usable even for $20/month.
OpenAI somehow went in the opposite direction. The limits are broken not only on the $20 plan, but even on the $100 and $200 plans. People are saying their limits sometimes doesn鈥檛 even last a full day.
2. Astra at $25/MTok output, ideally even less
The GPT-6 lineup used to be Astra, Sol, Terra, and Luna. Terra is gone now, and it feels like Sol was nerfed into what Terra used to be.
So realistically, Astra is the only true frontier model in the lineup.
And naturally, people want to use the strongest model, not pay for a subscription and then have to use the mid-tier one because the frontier model is too expensive.
3. Forced resets shouldn鈥檛 move your scheduled reset date
If you still have 80% of your limit left and your regular reset is in 2 days, getting a forced reset shouldn鈥檛 move the next one to 7 days from now.
You basically get downgraded for receiving a reset, because you lose those 2 days you had already planned around.
4. Remove the 5-hour limit from the $20 plan
It鈥檚 just annoying and makes usage much harder to plan, especially when you already have upcoming resets to think about and want to manage your remaining allowance around them
What OpenAI is likely to ship at DevDay:
- o - Grok Bot alternative (btw apparently based on OpenClaw)
- $500 plan with access to Ultrafast mode at 700 tok/s
- Astra 6.1 (not too sure about this one)
- OpenAI speaker (or maybe even two)
- an update to voice, images, or web chat
- 6 banked resets 馃憖