@swyx

–

24 April 2026 6h41

i think the most underated part of today's launch is not GPT 5.5 at all

Compute Infrastructure Costs Vastly Exceed AI Model Training Expenses

By

–

24 April 2026 2h32

perhaps interesting to think through that compute infra building is probably a 0.5-1T problem, but training, if cursor is all it takes, is just a 0.06B problem

Extracting and Animating Components from ChatGPT Codebase

By

–

24 April 2026 2h15

haha understandable. just wanted to learn how to do this but with more handholding. nw u already gave broad strokes. i think i just sunnl how exactly to extract and animate individual components, but i think for u the answer is “just open the chatgpt codebase lol”

Codex App Transformation: Atlas Integration Reshapes User Experience

By

–

24 April 2026 1h44

wow the codex app is unrecognizable… almost like it shouldve been Atlas the whole time https://t.co/PLTMJQwJmQ
— swyx 🇸🇬 (@swyx) 23 avril 2026

wow the codex app is unrecognizable… almost like it shouldve been Atlas the whole time

AI Researchers and Rocket Scientists Managing Burn

By

–

23 April 2026 23h02

ai researchers rocket scientists managing burn

CUDA Kernels and Custom Heuristic Routing Algorithms

By

–

23 April 2026 20h59

"custom heuristic algorithms" was the cuda kernels yeah? wasnt exactly clear in the wording or if this part is hinting at some kind of cool routing work

Pareto Frontiers: Extended Context and Improved Inference Speed

By

–

23 April 2026 20h14

looks like new Pareto frontiers across everything:
– Context: 400K context in Codex and a 1M in API
– API Pricing: $5/m input and $30/m output tokens.
– Codex improved its own inference speed 20% lol
– First generation co-designed with GB200 and GB300 NVL72
– 82.7% on… https://t.co/J5CL5fKmsq pic.twitter.com/nJG1vubSdN
— swyx 🇸🇬 (@swyx) 23 avril 2026

looks like new Pareto frontiers across everything: – Context: 400K context in Codex and a 1M in API
– API Pricing: $5/m input and $30/m output tokens. – Codex improved its own inference speed 20% lol – First generation co-designed with GB200 and GB300 NVL72 – 82.7% on

Industry dynamics and structural transformations analysis

By

–

23 April 2026 18h58

on the dynamics of what happens to an industry

GPT-Image-2-Thinking: Extended Generation Time, Enhanced Capabilities

By

–

23 April 2026 4h27

btw in talking to friends the best framing for how to discuss GPT-Image-2-Thinking taking multiple tens of mins for generation and being able to oneshot QR codes and diagrams and logos and foods and faces..

…is that Image-2 is a new Image model, Image-2-Thinking is a new Image… https://t.co/Bca7bY4kBb pic.twitter.com/5DVajV0RMV
— swyx 🇸🇬 (@swyx) 23 avril 2026

btw in talking to friends the best framing for how to discuss GPT-Image-2-Thinking taking multiple tens of mins for generation and being able to oneshot QR codes and diagrams and logos and foods and faces.. …is that Image-2 is a new Image model, Image-2-Thinking is a new Image