I’d like to request a visible usage meter for ChatGPT Live Voice, particularly when users have different time allowances for higher-capability Voice models and Live mini.

Recently, I was having a long Live Voice conversation and reached my higher-capability Voice usage limit. The conversation continued, but the underlying model changed. The difference was immediately noticeable to me: the conversational cadence, responsiveness, personality, and overall feel of the interaction changed enough that it suddenly felt like I was speaking with a different conversational partner.

What made that especially jarring was that I had no clear indication beforehand that my higher-capability Voice time was nearly exhausted. I initially didn’t know why the experience had suddenly changed.

I’m not necessarily asking for the usage limits themselves to be increased. I’m asking for better visibility and transparency around the limits that already exist.

I think Live Voice would benefit from showing:

  • Remaining higher-capability Voice time

  • The applicable rolling usage window

  • When additional usage begins becoming available again

  • A warning when there are perhaps 15 minutes and 5 minutes remaining

  • A clear notice before the conversation switches to a different Voice model

For example:

Live Voice — High
42 minutes remaining
Rolling 24-hour allowance
Additional usage begins becoming available at approximately 3:17 AM

And before a transition:

“Your higher-capability Voice allowance is almost exhausted. This conversation will continue using Live mini.”

I think this is particularly important for conversational AI because changing the underlying model can affect much more than raw capability. It can noticeably change the cadence, responsiveness, conversational style, and continuity of an ongoing interaction.

A model transition therefore isn’t necessarily comparable to something invisible happening in the background. If the conversational experience itself is about to change, users should be able to see that coming and plan around it.

A simple remaining-time meter and advance model-transition warning would make Live Voice significantly more predictable and transparent.

I’d also be interested to know whether other Live Voice users have noticed the same kind of abrupt change when their higher-capability allowance runs out.