The most brilliant people to me are the ones that can take the most complex information, data, task or jobs and explain it to the most basic intelligence people in a way they can understand it with ease. That’s how I can do most and learn most things, I reverse engineer it.

Guessing we’re getting another reset tomorrow since I haven’t seen anything new announced. Hopefully because I need all the usage I can get by Tuesday.

And now down to 60% after a full day of usage so we got better but still crazy how it can eat that much in less than a day. I’m only working on a single app and only working in 6.1 sol. I get it, 6.1 Terra is about to come out but I’m tired of having to choose models. They say “only use Astra on hard tasks” but what if the app you’re building is all a hard task? Apparently it is because when I try to use weaker models it’s almost as if it forgets all the rules and anything I previously taught it with skills. Dot has gotten dumber, my usage is cut in half, I’m having to use different model after different model just to build an app. They gave us 65,000 credits but I have 7000 left after having to use it for work since they cut the usage in half. I’ve been a heavy $200/mo user since they announced it. I’ve only tried Claude a little bit when fable was a trial and loved it but I’ve put so much work into my codex account that I’d hate to switch now. I’m getting tired of spending all this money to not be able to use it all week even. Don’t tell me they can bring back 5Hr limits because I need to be able to work for sometimes 10-12 hours a day. I would get it if I was on a lower plan but come on now you just made a plan for what 150% more a month just to get what I previously had and paid for? All these free resets are cute but I can’t trust my whole company to get usage to work all week so I see no choice but to go with the frontier team that is able to allow a team to work all week. I personally think open source wins because these companies are going to be our bottleneck.

This is the level of smart I aspire to be. Bro is out here altering the genome of the model lol. I need to focus my ADHD hyper focus super powers better because this is sick.

Why does it feel like my Dot gets dumber the more I use it?

Down to 43% of my codex usage after yesterday’s reset using sol 6.1 sol max only. Before they cut my usage in half I could’ve done this for 3 days straight to hit 40%. I think I’m gonna have to look into Claude on my personal account. I gave it some time and if it weren’t for @thsottiaux resetting damn near everyday this is almost un-usable now even on their premier “affordable” model. I haven’t touched Astra in like 2 weeks because I’ve been trying to stretch usage. Playing with Hermes and local models has me thinking that’s the real endgame, not these frontier models that can bottleneck your usage at any moment cutting it in half.

What I need @OpenAIDevs to do is just take the codex harness and put it in my Dot… why is that so hard?

I miss when it felt like I had unlimited usage on 6.1. Now, I do think @thsottiaux knows and understands how bad it is having our usage cut it half so I think they have something up their sleeve to make us whole again at the end of this 28 days but I really hope I’m right because 6.1 is really slow and I feel like I’m fighting it a ton compared to Astra.

I wish I had the option to take a cloud chat and make it local or vice versa take a local chat and make it cloud. This would be an awesome change @thsottiaux

For those of you that speak multiple languages and now use a different language as your primary, is it difficult for you not to subconsciously slip back into your old language? Also which language does your inner monologue speak?

Spending my night making my Dot train and starting a nightly refresher for it to update itself. I’m getting tired of the regressions.

I feel so out of place when I try to talk to anyone at all around me about Ai, what I’m building, how I’m building it, what I’m learning. They just look at me with a blank stare. I need more friends in tech.

Who is gonna be the first lab to open source a beast local model for the 4090 or is Qwen 4 27B going to be that far ahead?

Well I thought my Dot was almost unusable before but now it’s telling me it’s reaching capacity with every model. So not only can I not talk to it with voice because it disconnects but now it can’t send tasks to my desktop.

I get zero sleep now that I have Ai running constantly. I’m always thinking of something new I can try. I’m pretty sure I’m going to be able to hand off my RAG tool next week so I’m porting it to my desktop in order to test it on windows. I love Tailscale.

You’re all welcome for the reset by the way. Nobody can resist thee ole Oliver Twist. Thanks @thsottiaux

I’m either gonna be out here begging @thsottiaux for a reset like Oliver Twist this coming week or I’ll be relying heavily on Qwen 3.8 on Hermes to do my work. Not even 6.1 and dots can let me work all week anymore sadly.