Phil Chen

@thsottiaux Generate synthetic data to train better models!

With Codex the there is quite the gulf in load between peak and off-peak times, and we would like to achieve more of a smoother traffic pattern as that would be a more optimal use of our compute. We have ideas, but curious what you all think we should do? Would more usage during off-peak and surge multiplier during peak times make sense?

@michael_silver no room with Apple watch on one arm and whoop on the other

@philhchen Sir can I introduce you to, the wrist watch?

Jewelry for tech bros the biggest untapped market opportunity today

@javiercorrea yessir

@philhchen Should we get a pair?

design has fundamentally changed with @Paper MCP + agentic coding tools. so much alpha in this blog about AI-native design in 2026

@thsottiaux @thsottiaux codex UI is super laggy once I'm on 5 parallel agents in worktrees. switching between threads takes ~2s

What are we consistently getting wrong with codex that you wish we would improve / fix?

@tkkong 🫡

this week I've truly felt like an agent fleet commander: - codex for most coding tasks with multiple worktrees - claude+paper for designing and implementing UI company and product building in 2026 is fundamentally different

excited to coauthor this article with @amytam01. at Quadrillion we happen to think that research is the most important thing that we can do as people, and we want everyone to have that capability (1/2)

@LucasVHoutven @thsottiaux For me actually I had > 2x more progress in a few hours than I’ve ever had with a model

@philhchen Yeah I noticed this too the model just EATS through usage in no-time. Its is unusable! @thsottiaux please do something about this! We pay 200 per month and now we have to worry about this again! With 5.3 codex never had a worry in my life about usage!

I feel so proud. first time running into Codex rate limits on Pro. congrats to everyone who worked on 5.4 - it's a great model

anthropic leaks of 2026 are the new openai leaks of 2023

@ycombinator @androo_sy @DavidhNie congrats on launch!

Polymorph helps consumer & self-serve apps deepen engagement: they build live user profiles from in-app behavior and send custom messages at the right time and channel. Early customers already see 3.6x better reactivation rates. Congrats on the launch, @androo_sy, @DavidhNie, and Manas! ycombinator.com/launches/PZ3-p…

@OfficialLoganK @NanoBanana Congrats on the launch!

Say hello to Nano Banana 2, our best image generation and editing model! 🍌 You can access Nano Banana 2 through AI Studio and the Gemini API under the name Gemini 3.1 Flash Image. We are also introducing new resolutions (lower cost) and tools like Image Search!

@arampell I actually went through this list and didn’t like any Tolkien nouns that weren’t already taken :/

In case you're looking for a startup name, here's a list of nouns (people, place, or thing) from Lord of the Rings, whether it's evil/good/neutral, and whether it's taken by an existing company This is clear proof that the market is in the first inning!

@ankesh_anand Great work, looking forward to more!

great article from phil, flash can hold its own against other frontier models :)

@dkundel Yeah renaming from slash command but low priority relative to forking

@philhchen Steering can be enabled in the settings since yesterday :) rename is currently only by right clicking the thread in the sidebar. Do you want to be able to do it from composer or explicitly using the slash command. Fork is a common request that is on the list

codex team really cooked with the new UI and model. I keep wondering if I'm just bad at tmux but the UI makes codex way more enjoyable to use