quite funny @mediastorm_en good job!!
Ron Paul retweeted
Study hard what interests you the most in the most undisciplined, irreverent, and original manner possible.
Ron Paul retweeted
You need to develop a fierce intolerance of badness of any kind, regardless of its severity. #principleoftheday
Honestly, cutting Astra’s price in half might be one of the best moves you could make. The model is clearly capable—the bigger problem is that the current pricing makes it hard to justify for everyday use.
Hi,
Tomorrow we are re-opening the Pro $200 subscriptions to new subscribers, but together with it we are also changing how we calculate the usage for it. In effect, if you do the math, it will net out at half the dollar in API spend compared to the old Pro $200 plan.
Now that it's said, let me explain why this is happening and why you will still get more work done than if you were on the Pro $200 subscription one month ago.
(a) We didn't want to compromise in other ways and are committing to not reintroducing the 5h limit, so that you can fully use the weekly usage when you want.
(b) On the subscription, we guarantee that over time you always get more work done and with an increasing level of quality. This means that you will continue to get more value per dollar spent as a result of models getting more efficient and us passing down the improvements in the form of API price reductions.
(c) We don't want to put an incentive on ourselves to artificially inflate the API list prices to make it look like you are getting a lot (and workaround it through discounts, etc). Instead we want to continue to both rapidly reduce prices and increase capabilities of models on the API. This week we introduced GPT-6 Sol and GPT-6 Luna at 50% of their previous price. Over time, we see prices go low enough that it makes sense for most to buy usage as needed without there being a significant gap between what you get in a subscription and what you get in the API for a dollar spent.
(d) Tomorrow, we are adding more things to the subscription that won't draw on the usage, I won't reveal what that is yet.
I wanted to be transparent before all the big announcements tomorrow. Lots of new exciting things are coming to the subscriptions that will make it super compelling, but I wanted to make sure to share this change ahead of time so you can all understand it before we shower you with good news.
Codexingly,
Tibo
I can’t help wondering whether the “safety delay” is only part of the story. If Astra 6.1 launches and fails to clearly outperform Opus 5.5—or even Sonnet 5.5—that would be a pretty awkward outcome after all the anticipation.
Either way, I’m still rooting for OpenAI to ship something genuinely strong. Astra has impressive capabilities, but its cost makes it hard to justify for everyday use. What I really want is Astra-level capability at a price that actually makes sense in production.
Tried GPT-6 Astra on medium after seeing a lot of people say it’s enough, but it updated the code and missed the corresponding docs. That’s not great for long-term project maintenance, so I’m trying high next to see if it stays more consistent.
@thsottiaux @jxnlco @victornunez
I think there’s a worktree-scoping bug in the Codex app. If a project’s CWD is inside a linked Git worktree, threads can discover two copies of the same skill: one from the active worktree and another from the main worktree. This makes it unclear which version will actually be used. Skill resolution should likely stay within the project’s current worktree. @Dimillian @pvncher @reach_vb @ajambrosino
It’s finally here! My Codex Micro has completed its long journey across the ocean and landed on my desk. As a longtime Codex fan, I couldn’t be happier to finally have one. Absolutely worth the wait — can’t wait to put it to work! ⌨️🤖✨
@Dimillian @nickbaumann_ @thsottiaux
#Codex #CodexMicro
ChatGPT Work has become my thinking and execution partner. It helps me turn rough ideas into clear plans, navigate complex codebases, draft code and docs, run checks, and keep long projects moving—saving me hours while I stay in control. #ChatGPTWork
Codex Mobile issue: when a repo has multiple Git worktrees, the same relative path (e.g. A) exists in each one. I can create a thread with A in the main worktree as its CWD, but I can’t choose A from another existing worktree. Please support worktree-aware CWD selection. @Dimillian
Codex now has a max gear in the model-strength enum.
You can see it by dumping the codex_app.* tool-call schemas. When an enum grows a new ceiling, something is usually about to hit it. 👀
@thsottiaux @ajambrosino @nickbaumann_ @dkundel @Dimillian
#Codex #OpenAI #GPT5p6
Nothing says “IPO season” like turning internal productivity metrics into a civilization-scale prophecy.
Our internal data shows Claude is accelerating AI development—a possible path to recursive self-improvement, or AI autonomously building a more capable successor.
It’s happening faster than we thought, and the implications deserve greater attention. anthropic.com/institute/recu…
I think I found the interesting part behind the new Codex behavior: thread management is becoming toolable.
The newer build exposes more codex_app calls around threads — sending messages, renaming threads, and manipulating thread state. That’s a big shift from my previous approach of reading session JSONL files to reconstruct context and execution state.
Once threads are programmable objects, you can imagine a lot more: skills that spawn or continue threads, workflows that hand off work across projects/worktrees, and tooling that orchestrates Codex instead of prompting Codex to orchestrate itself.
This feels like a much more powerful control plane. 🧵
