2026年2月归档
Personal view. On the DoW deal, time an…
Personal view. On the DoW deal, time and time again I witness how OpenAI operates thoughtfully and diplomatically when i
引用：Codex is now starting to be associat…
Codex is now starting to be associated to speed. Excited for the team to keep building here, there is a lot more we can
引用：Truthfulness is useful for engineering…
Truthfulness is useful for engineering & research (and hopefully beyond). But maybe we are naive :) "Unlike the [redac
引用：Truthfulness is useful for engineering…
Truthfulness is useful for engineering & research (and hopefully beyond). But maybe we are naive :)
Since we released it in the API, I'm…
Since we released it in the API, I'm consistently hearing from folks at meetups that GPT-5.3-Codex is the one model to u
We have invited our first (small) ba…
We have invited our first (small) batch from the waitlist to try the Codex app on windows. Hope to quickly expand from
引用：Major upgrades landed in the Codex C…
Major upgrades landed in the Codex CLI, upgrade to v0.105 $ npm i -g @openai/codex@latest Syntax-highlighting, dictate
Codex Whendows
Codex Whendows
引用：I have a solution
I have a solution
引用：Yesterday Rohan asked me if we have a w…
Yesterday Rohan asked me if we have a written down mission. I said no, but that we are building THE agent that will powe
The good thing about working at an AI l…
The good thing about working at an AI lab is that sometimes you get to see a very cool chart. The bad thing is that you
We brought more compute online in Fe…
We brought more compute online in February to sustain Codex demand than we did in the entire period since its inception.
引用：Only the spark model is served throu…
Only the spark model is served through Cerebras, all speed optimizations for GPT-5.3-Codex are something different and t
There are models that do things and the…
There are models that do things and then there are models that pretend to do things, but are nice about it. Beware
引用：The claw must remain open
The claw must remain open
We’ve made GPT-5.3-Codex-Spark about…
We’ve made GPT-5.3-Codex-Spark about 30% faster. It is now serving at over 1200 tokens per second. More to come on speed
dream it, codex it, watch it, yeet i…
dream it, codex it, watch it, yeet it, try it, tune it, ship it, rename it technologic
引用：Our codex offsite left a deep impres…
Our codex offsite left a deep impression on me. I am beyond excited for what the next 10 or so weeks will bring and I t
I am increasingly asked during candi…
I am increasingly asked during candidate interviews how much dedicated inference compute they will have to build with Co
You only need your ChatGPT subscript…
You only need your ChatGPT subscription to use Codex. It’s already included and even the plus subscription has very gene