Jonathan lora

[Jonathan lora @Jonathanlora17]

@dtzy_88 A few weeks ago I unexpectedly ran out of usage. It was so frustrating that after the reset, I built an automation where Sol would plan the task and send detailed instructions to a Luna chat, while Luna handled the actual execution. By the way I have pro x20.

[Jonathan lora @Jonathanlora17]

@OpenAIDevs Windows 💔😭

[Jonathan lora @Jonathanlora17]

@zswaff @thsottiaux 😂😂😂

[Jonathan lora @Jonathanlora17]

@YigitTabel @SebastienBubeck I had a feeling you were joking, but I decided to be helpful just in case 😭😂

[Jonathan lora @Jonathanlora17]

@daniel_mac8 But do you think it could launch this week? I thought it was scheduled for September

[Jonathan lora @Jonathanlora17]

@YigitTabel @SebastienBubeck Astra is OpenAI’s next major model 👀 Still unreleased though.

[Jonathan lora @Jonathanlora17]

@DeeperDev @xt0n1t3ch @thsottiaux At least they give you resets. On Anthropic, they wouldn't give you anything, and you'd burn through your usage even faster

[Jonathan lora @Jonathanlora17]

@0xShoosh @thsottiaux @ajambrosino @embirico Quick update: I spoke too soon. After the latest update, Codex worked normally at first, but after about 20–40 minutes the instability came back and feels pretty much the same as before. I’ll keep testing it and see how it behaves.

[Jonathan lora @Jonathanlora17]

@loganpendragonm @0xShoosh @thsottiaux @ajambrosino @embirico I think the issue is fixed.

[Jonathan lora @Jonathanlora17]

@0xShoosh @thsottiaux @ajambrosino @embirico I think it's fixed

[Jonathan lora @Jonathanlora17]

@ReinhardRT @OpenAI @ChatGPT @thsottiaux I think it's already resolved.

[Jonathan lora @Jonathanlora17]

@ReinhardRT @OpenAI @ChatGPT @thsottiaux Me too 😔

[Jonathan lora @Jonathanlora17]

@thsottiaux I haven’t really used Google Docs for this, but lately I’ve been using Codex with Obsidian a lot. Whenever I need documentation, an audit, or some notes, I just ask Codex to write it and organize it in my vault, and I love how it’s slowly becoming my own knowledge library.

[Jonathan lora @Jonathanlora17]

@Aongkou @thsottiaux What’s worked for me is using GPT-5.6 Sol Ultra mainly as a planner/supervisor: it analyzes the task and delegates the actual implementation to another chat running GPT-5.6 Luna Max. That’s basically how I’ve been able to keep my usage under control.

[Jonathan lora @Jonathanlora17]

@Aongkou @thsottiaux Same here. I’m on Pro x20, and for the last 1–2 weeks I’ve definitely felt like my quota is being consumed much faster, especially when using GPT-5.6 Sol Ultra.

[Jonathan lora @Jonathanlora17]

@Aongkou @thsottiaux I also avoid 1.5x/Fast mode now. Standard speed is much more manageable, but at 1.5x it feels like the quota just disappears. I honestly wouldn’t recommend using Fast for long sessions right now

[Jonathan lora @Jonathanlora17]

@violettspark

[Jonathan lora @Jonathanlora17]

@thsottiaux The app still has the occasional bug especially on Windows, but they tend to get fixed quickly, and it feels like the product keeps improving constantly.
And honestly, those Tuesday and Thursday updates have become a little addictive 😂 I always want to see what they shipped next

[Jonathan lora @Jonathanlora17]

@thsottiaux What I like most about Codex is how complete the experience has become: goals, the built-in browser, subagents, and now delegating tasks from one chat to another.
For me, it strikes a great balance between model intelligence, efficiency, and reliability when writing code