GPT‑5.3‑Codex‑Spark5.3
An ultra-fast model for real-time coding in Codex
Loading page
An ultra-fast model for real-time coding in Codex
GPT-5.3-Codex-Spark is an ultra-fast real-time coding AI model now available as a research preview for ChatGPT Pro users. With a 128k token context window and over 1000 tokens per second generation speed, it enables truly responsive, interactive collaboration — letting you interrupt, redirect, and iterate with near-instant feedback. Tuned for speed, Spark delivers targeted edits and code refinement without running tests unless you ask.