BluePi @BluePi_In

I have four ChatGPT Pro accounts. Three are completely out of GPT 6 Astra usage. The last one will be gone today. $600 a month for three days of Codex. I cannot buy more even if I wanted to. Pro subscriptions are closed. This is ridiculous. @OpenAI give us a reset today. Your paying customers are locked out of the best model you have ever built. How is your GPT 6 Astra usage looking?

🚨 Wait… WHAT happened to GPT-Astra after launch? 👀 Day 1: HOLY SHIT. THIS IS AGI. Day 2: This is the smartest model, I might never use another model again. Day 4: Wait… why is it struggling with things it crushed before? 😭 Day 6: Nah bro, something feels different. OpenAI nerfed it. Day 7: Opens GPT-5.6 Sol 💀 The Astra honeymoon ended QUICK. 👀😂

🚨 Tibo says OpenAI made NO changes to Astra since launch. Then why does Astra suddenly feel so much worse? 👀 - Same tasks. - Same prompts. - Wildly different results. Users are seeing less reasoning, more loops, and straight-up clueless behavior. So what changed? If Astra really wasn't nerfed, something else absolutely fucking changed. 👀

🚨 astra is nerfed maybe they have fucked up the juice /effort value before vs after in max < it took 9m 08s before and now 4m 48s> basically half the time and thinking effort

keyboard_arrow_left Previous keyboard_arrow_right Next

It feels like OpenAI has nerfed GPT Astra. just look at this difference. the left one was built at launch of GPT Astra and the right one was built right now.

@thsottiaux Thanks for the sketchy Astra „optimization“. You guys couldn’t really wait any longer for the nerfing, could you? Same circle at every new OpenAI model release over and over until the model gets unusable just like you guys did with gpt-5.6-sol

There’s strong evidence that some GPT-6 Astra users have been getting a degraded and inconsistent experience since roughly Sep 6–9. I’ve been digging through OpenAI’s GitHub, and the broader ChatGPT/Codex stack has also been having problems: capacity errors, context/compaction issues, quota/reset incidents, and even a merged fix involving reasoning-effort state persisting incorrectly through compaction. None of this proves Astra itself was “nerfed,” but something in the stack has clearly been off. I’ll leave the links to the investigation trail in the comments below, including the main user reports and relevant OpenAI Codex changes I found.

@AnthropicAI So our data isn't really private as we thought?

GPT 6 Astra limits are TERRIBLE on a $200 ChatGPT Pro subscription. I just blew through my entire week in about 4 hours of work. OpenAI keeps doing these resets that are just prolonging the inevitable. Fix the limits @OpenAI .

GPT 6 Astra is draining Codex subscriptions. GPT Astra is the most token efficient model ever built and it still eats a full week in a day. The only thing keeping Codex usable is OpenAI handing out resets every 48 hours. Without them subscribers get almost nothing. Fix the limits @OpenAI.

Yesterday evening I reached 0%. Today I used one of my banked resets and yet I've used 85% already 😥 @thsottiaux will you please rescue me? 👉👈

You know it's serious when Pro 20x people start complaining😂😂😂 WHILE USING ASTRA LIGHT, OpenAI really need to fix both plus and pro Plan limits, there's no point in raising price and locking everything for Pro. If Plus users can't use it, then Pro users will also not be able to use it properly.

Kol Tregaskes @koltregaskes

3 weeks ago

Wow does Astra eat your tokens. I don't know where these cost efficiencies are but I've been using Astra Light for only a day and I'm basically out of my weekly usage on Pro x20 plan. Grok and Claude it is then for the rest of the week.

123 27 757 64K 31

Wow does Astra eat your tokens. I don't know where these cost efficiencies are but I've been using Astra Light for only a day and I'm basically out of my weekly usage on Pro x20 plan. Grok and Claude it is then for the rest of the week.

The only thing you forgot to mention is that even on Low reasoning, Astra burns through limits faster than Sol on Max!

Tibo @thsottiaux

2 weeks ago

To calibrate you all on which reasoning effort to use for Astra, know that GPT-6 Astra on low performs better than GPT-5.6 Sol on high. If you were using high reasoning efforts with Sol and were happy, I suggest you move down to low or medium for Astra.

2K 1K 23K 3.3M 3K

Banked resets might not be as generous as they look, Do you get this? Here's the comparison of same 14% allowance drain before vs after a banked reset: Before: ~115M tokens After: ~64M tokens That’s almost 50% less usage for the same 14% drain 😭 The concerning part? Using the banked reset also pushes the next weekly reset another 7 days out. Obviously we can’t see what’s happening server-side, so this isn’t definitive proof, but if the analysis holds, a banked reset could be giving significantly less allowance than a normal weekly reset while still restarting the weekly clock.

WOW: I used up a 20x plan in one day. Anyone else?

What's the point of all that Astra power if a $200 subscription only lasts you a single day?

@elshayib_ Yep, somehow math is not mathing 🤔 I also tried and what I am usually doing with Sol Max, draining 50% of my 5h limit, with Astra Medium I drained around 120% (1x5h limit + another round from another 5h limit). And honestly, result was slightly better but not that much.

OpenAI is playing everyone right now how the hell is Astra more than 3 times more efficient per task and yet it consumes usage a lot faster than running Sol. I understand you want better margins but FFs I feel like I'm getting Finessed straight up, i love Astra it's great but I think I'm not the only one running out of usage so fast like that. Recurrent depth was all about efficiency and it's doing great and the model is clearly more efficient but we got nothing out of that as customers wtf, at least give us the same usage as Sol.

Does GPT 6 Astra usage drain fast with Codex?