Codex adds next-message predictions for personal Pro users
Your move: Tab accepts a suggestion into the composer; you still review and send it
Read more
OpenAI announced composer predictions in Codex on October 9. After Codex finishes a response, it can suggest a complete next message based on the current thread. Pressing Tab adds the suggestion to the composer, where you can review or edit it before sending. The beta is for adults aged 18 and older on personal ChatGPT Pro plans in supported regions, using local or SSH threads in the Codex desktop app with GPT-6 Astra or GPT-6.1 Sol. Predictions are on by default for eligible users and can be turned off under Settings > General > Composer > Show predictions.
Why it matters
This can make steering a coding task quicker, but it does not let Codex continue work by accepting or sending follow-ups on its own. Predictions use context already in the current thread; they do not independently look up memories or connected apps. During the beta, generating predictions costs no extra and does not consume Codex usage limits or credits. Messages you actually send, including accepted predictions, follow normal usage limits and billing. The October 9 announcement introduces the feature as a beta, not access for every Codex user.