Tools ·
GPT-5-Codex tunes GPT-5 for agentic coding, thinking longer when tasks demand it
OpenAI released GPT-5-Codex, a version of GPT-5 optimised for software engineering in Codex, which adjusts how long it thinks to suit task complexity and can work independently for extended periods. The update also added code review and unified Codex across the CLI, an IDE extension, the cloud and GitHub. For practitioners, it made one agent and model usable consistently from the local terminal through to pull-request review.
Read the full story at OpenAI ↗Like, save or discuss on Buzz
