Hi OpenAI team,
I’d like to suggest adding advanced speed modes in Codex, especially now that the usage reset feature exists.
Right now, the speed menu offers Standard and Fast / 1.5x. It would be very useful to have optional higher-speed modes such as:
- Standard
- Fast / 1.5x
- Very Fast / x3
- Turbo / x6
- Max / x12
The goal would be to use these modes for long autonomous Codex sessions: audits, log analysis, file exploration, safe refactors, tests, documentation, and multi-step implementation work.
This would work especially well when combined with the new usage reset feature. If a user has a reset available, Codex could let them intentionally choose a higher-speed mode for a long work session, with clear visibility on expected usage and remaining resets. For example: “Use Turbo x6 for this session. This may consume usage faster. You have 1 reset available.”
I understand that higher speed would likely increase usage/cost. That’s fine as long as it’s clearly shown and opt-in.
Suggested safeguards:
- Disabled by default
- Clear confirmation before x6 or x12
- Visible warning about increased usage
- Show remaining resets before starting a high-speed session
- Let the user combine high-speed mode with a reset intentionally
- Automatic pause before dangerous actions: production restarts, destructive commands, database migrations, billing changes, process kills, deletions
- Regular checkpoints with summary, files changed, tests run, and rollback info
- Automatic fallback to normal speed if errors repeat or if the task becomes ambiguous
Why this matters:
For users who use Codex as their main coding agent on long operational tasks, 1.5x is useful but still limited. Optional x3/x6/x12 modes, combined with usage resets, would make Codex much more effective for long-running autonomous work, especially when the user is willing to spend more usage in exchange for faster execution.
Thanks!