GPT 5.3 Codex Spark is a smaller, ultra fast version of GPT 5.3 Codex built for real time coding inside the Codex app, CLI, and VS Code. It runs on Cerebras hardware with more than 1000 tokens per second, has a 128k context window, and is tuned for quick, targeted edits rather than long agent runs. It is rolling out as a research preview to ChatGPT Pro users with separate rate limits while OpenAI scales up capacity.
π₯ Our Take: There has been a lot of talk about agents that think for minutes or hours before you see anything. This update is clearly for the other kind of developer, the one who wants to nudge code in small steps, interrupt mid stream, and keep moving without staring at a spinner. I am very interested to see how many people keep this as their everyday default and only reach for the heavier model when they really have to.