GPT 5.6 Sol ultra is continuing to impress
Set a new thread w/goal to fully create, implement and launch @Houspensive website with a clear CTA.
It’s been running for over 4 hours - just got notified my @Cloudflare and @GoDaddy name servers changed….
6 months ago this wasn’t possible.
Computer Use + Plugins and providing clear direction does wonders
GPT 5.6 Sol launched, started a brand new thread and goal, but it seems everyone is doing the same as the servers are at capacity already 😂 @thsottiaux this deserves another reset for the one time hahahaha
Isn’t the whole point of having Memories enabled, while using a project directory where the Memories and Agents .MD lives so that it can be referenced with a new thread in the same project?
I’m currently using a thread for a singular prompts request, then once it finishes I open a new thread.
This was my issue at the beginning too…. I would find threads would take forever after I’ve used them continuously for days. That’s why I use separate threads now.
Don’t memories within a project hierarchy stay relevant and saved so future threads can read them from their same project?
@thsottiaux I’m sure you’ve had mass amounts of people tweeting at you with this same complaint or similar.
Any update you could provide? Or, are all us codex users delusional with these statements we’re making regarding usage and limits being way lower than before?
Any update would be great my dude
Not sure what has happened to Codex, but I am lucky if I can get 2-3 prompts completed before my session usage is maxed out.
To get around this, I started creating way larger/more detailed prompts, and assigning the /goal to every prompt.
/goal seems to continue running even when I am out of usage limits, it will finish its “goal” which is pretty awesome.
Anyways, with a reset every 5 hours, I typically can get 2-3 goals a day “done” or ran.
Hoping they optimize the token usage/session limits before GPT-5.6 is release. (Or whatever the heck they decide to call the next version.)
@simpsoka Yes. The answer is a huge yes. We need the whole usage limits issue reconciled. I used to spend 8 hours a day sending 50 prompts and was never hitting limits. Today was a record for how fast it disappeared…
Not sure what has happened to Codex, but I am lucky if I can get 2-3 prompts completed before my session usage is maxed out.
To get around this, I started creating way larger/more detailed prompts, and assigning the /goal to every prompt.
/goal seems to continue running even when I am out of usage limits, it will finish its “goal” which is pretty awesome.
Anyways, with a reset every 5 hours, I typically can get 2-3 goals a day “done” or ran.
Hoping they optimize the token usage/session limits before GPT-5.6 is release. (Or whatever the heck they decide to call the next version.)
If I’m today’s giveaway winner then yes, otherwise, loops only work well if you’re on the pro plan with way higher limits / usage.
I just used /fast with 5.5 Ex High reasoning on my Plus plan and ate 50% of my weekly limit and destroyed my 5-hour limit in a single prompt..
Loops are for Pro use only by this standard.
So like, when Montreal wins the cup this year, yall paying up or what?
The slander on the timeline is insane.
Habs vs Knights Final
Habs win in 5
Potentially 6 with a flukey goal in the 5th game by marner to force 6.