log404 @log404tv - Twitter Profile

Just me who find this policy puzzling? Free users get infinite usage via Luna, but when a Pro user runs out of credits, they don't even have the option to use Luna for free.

76

2K

36

140

269K

log404 @log404tv

10 days ago

@rezoundous Terra Xhigh

0

34

log404 @log404tv

11 days ago

@thsottiaux @AbdoKerdawy Sol hallucinations and focus loss in codex because of this makes it very hard to code on giant code bases or connecting different things.

0

4

0

1K

log404tv retweeted

Ishu Agrawal

@ishuagra02

11 days ago

I actually use Codex a lot... more than most people asking for resets in the replies. And it is with my heavy usage that I know that Codex limits are just not as good as they used to be. With GPT 5.5, I had a $100 plan and it'd be impossible for me to meet my daily and weekly limits. I'd have Codex working on 2 projects concurrently, running code reviews, using image gen, and controlling my browser. And I still would have usage leftover. Now, with GPT 5.6, an entire week's of usage can be burned in a couple hours with GPT 5.6 Sol Ultra on the same $100 plan. This is absolutely absurd. It was so bad that OpenAI had to investigate this for weeks once they received complaints from countless users. And Tibo confirmed that GPT 5.6 Sol used more tokens than GPT 5.5 at the same reasoning. They've made some improvements to their tool calls that should have resulted in 18% higher usage with Sol. And I'm grateful for the team's effort. But it still feels worse than GPT 5.5 did. And I know I am far from the only one who thinks this. I still love my ChatGPT subscription. I love the no weekly limits. I love its technical excellence at coding. I love that there's no gatekeeping of models. And I vastly prefer it to Claude's. But I won't pretend that limits have not felt worse with the introduction of GPT 5.6, despite OpenAI's claims that it's 50% more efficient in agentic coding. And those were largely masked by the frequent resets we received every 2-3 days. As OpenAI begins monetizing resets, free resets will be less frequent. Which can effectively reduce overall limits by up to 57%. Today, the $100 plan feels like the minimum to get any good usage out of Sol. Which GPT 5.4 and GPT 5.5 didn't desperately need. We're gradually treading an upward trajectory of how much money we need to spend to get the equivalent usage across consecutive releases. I'm really hoping GPT 6 or Astra doesn't represent another spike in overall usage, such that even the $100 plan won't feel sufficient.

ishuagra02's tweet photo. I actually use Codex a lot... more than most people asking for resets in the replies. And it is with my heavy usage that I know that Codex limits are just not as good as they used to be.

With GPT 5.5, I had a $100 plan and it'd be impossible for me to meet my daily and weekly limits. I'd have Codex working on 2 projects concurrently, running code reviews, using image gen, and controlling my browser. And I still would have usage leftover.

Now, with GPT 5.6, an entire week's of usage can be burned in a couple hours with GPT 5.6 Sol Ultra on the same $100 plan. This is absolutely absurd.

It was so bad that OpenAI had to investigate this for weeks once they received complaints from countless users. And Tibo confirmed that GPT 5.6 Sol used more tokens than GPT 5.5 at the same reasoning. They've made some improvements to their tool calls that should have resulted in 18% higher usage with Sol. And I'm grateful for the team's effort. But it still feels worse than GPT 5.5 did. And I know I am far from the only one who thinks this.

I still love my ChatGPT subscription.
I love the no weekly limits.
I love its technical excellence at coding.
I love that there's no gatekeeping of models.
And I vastly prefer it to Claude's.

But I won't pretend that limits have not felt worse with the introduction of GPT 5.6, despite OpenAI's claims that it's 50% more efficient in agentic coding.

And those were largely masked by the frequent resets we received every 2-3 days. As OpenAI begins monetizing resets, free resets will be less frequent. Which can effectively reduce overall limits by up to 57%.

Today, the $100 plan feels like the minimum to get any good usage out of Sol. Which GPT 5.4 and GPT 5.5 didn't desperately need. We're gradually treading an upward trajectory of how much money we need to spend to get the equivalent usage across consecutive releases.

I'm really hoping GPT 6 or Astra doesn't represent another spike in overall usage, such that even the $100 plan won't feel sufficient.

143

1K

45

79

127K

log404 @log404tv

11 days ago

@thsottiaux @ishuagra02 As an active sub for going on 3 months now limits before the constant resets were probably 10x longer than they are now for just the 5 hour window

0

3

0

826

log404 @log404tv

11 days ago

@pvncher How about their issue with loosing context how does it handle this?

0

208

log404 @log404tv

12 days ago

@ziwenxu_ Love this repo waiting on windows 🙏

1

0

31

log404 @log404tv

12 days ago

@elonmusk Can you improve the overall credit amount for subs to make it more affordable for us? 🙏

1

3

0

34

log404 @log404tv

13 days ago

@ziwenxu_ I heard some say that cursor has more than groks subs but maybe x has the best sub very confusing set of subs

0

13

log404 @log404tv

13 days ago

@ziwenxu_ How's the usage on that?

1

0

23

log404 @log404tv

13 days ago

@maria_rcks @sama did july to us sam?

0

2

0

162

log404tv retweeted

Elon Musk

@elonmusk

13 days ago

@SmokeAwayyy Specialists will be far more efficient

208

3K

150

159

175K

log404 @log404tv

13 days ago

@thsottiaux Can we get a little refresher after tomorrow? I blew through 40% of my limits with Sol on a very very simple fix trying to understand why it got stuck so long. 🙏

0

11