ChatGPT needs a visible conversation-capacity indicator for every chat.
The indicator should show:
- The approximate percentage of the current conversation context already used.
- The approximate remaining conversation capacity.
- Progressive warnings at approximately 70%, 85%, 95% and 100%.
- A clear distinction between:
- conversation/context capacity;
- account-wide message or model usage limits;
- ChatGPT Work, agent, research and tool-processing limits.
- The currently selected model and reasoning level.
- Whether the chat is operating in normal Chat mode or Work/agent mode.
- A warning before a chat is moved into a different workflow that cannot later be returned to normal Chat.
- An automatic handover/export option before the chat reaches its limit.
The handover should preserve the project name, chat name, active instructions, decisions, workflow rules, files and sources, current status, last completed action and exact next action.
This feature is important because users currently have no dependable warning that a long-running project chat is approaching its limit. Users may only discover the problem when the chat suddenly becomes unavailable or loses continuity.
The meter may be an estimate because hidden instructions, tools, reasoning and attachments affect token usage, but an approximate warning would still be extremely valuable.
Please record this as product feedback for the ChatGPT product team.