Extend banked Codex reset expiration beyond 30 days

What feature would you like to see?

Please extend the expiration window for banked Codex usage resets.

If banked resets currently expire after about 30 days, that window is too short for many users. A banked reset is most useful when it can cover occasional heavy workloads, vacations, travel, sick days, or periods where the user simply does not need Codex heavily that month.

A better policy would be one of:

• no expiration while the subscription remains active, or
• a much longer expiration window, such as 12 months, or
• at minimum, a grace window far longer than 30 days.

Why this matters

A 30-day expiration makes banked resets feel easy to lose accidentally. Users may receive a reset during a period when they cannot use Codex heavily, then lose it before the next real burst of work.

This weakens the main value of reset banking: smoothing uneven usage over time.

Expected behavior

Banked resets should remain available long enough for users to plan around them.

For example:

Banked resets: 1 available
Expires: Jun 18, 2027

or, if no expiration is used:

Banked resets: 1 available
No expiration while subscription is active

Related issues

• #28161 asks to show expiration dates for available resets.
• #28805 asks to show reset count, expiration, eligibility, and reset events in usage surfaces.
• This issue is specifically about the expiration policy itself being too short.