Sherwin Wu@sherwinwuChatGPT for Excel is here! GPT-5.4 is shockingly good at performing Excel manipulations. In particular, it's been impressive at handling work when thrown into complex existing spreadsheets. Available for Plus, Pro, Enterprise, Business, and Edu users! t.coOpens with an observation
Sherwin Wu@sherwinwuMy top use case for GPT-4o native image output right now: using it to imagine home reno projects in our new place. i.e. look at this potential built-in bookshelf / reading bench — rendered in one shot!Opens with an observation
Sherwin Wu@sherwinwuWe quietly launched the Symphony repo on Github last month, and it's already accumulated 15.5k stars! Excited to share this post that dives into it even deeper: a library that lets you use Codex to orchestrate work normally done by teams of engineers. t.coOpens with an observation
Sherwin Wu@sherwinwuEveryone's talking about the duplex architecture and the truly "live" interactions with GPT-Live. But the part that I'm most excited about it just how much _smarter_ this model is than other voice models. It can delegate to our other models to do work on its behalf!Opens with an observation
Sherwin Wu@sherwinwuWe're now in a world where you can get 54.4% SWE-Bench PRO performance, and 60% (!) T-Bench 2.0 performance: - at 2x the speed of GPT-5-mini - for 7.5x to 10x cheaper than GPT-5.4 Wild!Opens with an announcement
Sherwin Wu@sherwinwuBest part of launching o3-mini in the API: - available to our broadest user base yet: tiers 3 and up - comes with almost all developer features enabled - cheapest reasoning model yet - smarter than even o1 at times (at reasoning_effort: high)Opens with an observation