Feature request: show remaining GPT-6 Pro / Sol Pro Chat message quota in Usage and model picker

What variant of Codex are you using?

ChatGPT Desktop (the unified Chat / Work / Codex app) and ChatGPT Web. This request is specifically about regular Chat quota visibility alongside the existing Work/Codex usage surfaces.

What feature would you like to see?

Please expose an authoritative used / remaining / reset counter for fixed-count regular Chat model allowances, both in Settings → Usage and, preferably, as a compact indicator in the model picker.

OpenAI currently documents a fixed regular-Chat allowance for GPT-6 Pro and GPT-5.6 Sol Pro. For example, on Pro $100 the documented allowance is 50 messages per week shared across GPT-6 Pro and GPT-5.6 Sol Pro, and the same documentation explicitly states that ChatGPT Work and Codex have separate allowances:

https://help.openai.com/en/articles/20001354-gpt-56-in-chatgpt

The problem is not the size of that allowance. The problem is that regular Chat does not provide a documented authoritative numeric counter showing how many of those messages have been consumed or remain. A user can therefore know that a hard, discrete weekly entitlement exists, yet cannot reliably determine whether they have, for example, 40 messages remaining or 4 without maintaining their own tally.

This is especially confusing because Work/Codex already expose substantially richer usage state, while their allowance is separate from regular Chat.

Suggested canonical Usage view

REGULAR CHAT
GPT-6 Pro / GPT-5.6 Sol Pro
34 / 50 messages remaining
Used:       16
Remaining:  34
Resets:     17 Sep 2026, 11:32
Shared by:  GPT-6 Pro + GPT-5.6 Sol Pro

WORK / CODEX
5-hour:     67% remaining
Weekly:     82% remaining

The values above are illustrative only.

Suggested model-picker treatment

A compact line would be sufficient:

GPT-6 Pro · 34 / 50 remaining · resets 17 Sep

Acceptance criteria

For every hard-capped, fixed-count regular Chat allowance, the product should expose:

• total allowance;
• amount used;
• amount remaining;
• exact reset date/time (and timezone or local-time presentation);
• which models share the same pool;
• a clear distinction between regular Chat and Work/Codex pools;
• the same authoritative account state across web, desktop and mobile clients.

Settings → Usage should be the canonical detailed view. The model picker can provide the lightweight at-a-glance view.

This should use the server-authoritative account state, not a client-side estimate or locally reconstructed message count.

Why this is materially different from compute-weighted usage requests

This particular allowance is defined by OpenAI as a discrete message count. The service necessarily has to track enough server-side state to enforce that count and its reset window. Exposing that existing state is therefore much simpler and more precise than asking a client to estimate remaining compute-weighted usage.

A hard paid entitlement should be inspectable before it is exhausted, not only discoverable when further use is refused.

Additional information

This request is related to, but intentionally narrower than, existing quota-visibility work:

• #24182 — persistent 5-hour / weekly usage display in the ChatGPT/Codex app;
• #29647 — explicit remaining quota and reset timing in /usage;
• #35166 — visibility of the Work/Codex shared allowance from ChatGPT Web;
• #41617 — auditable per-feature usage attribution.

Those issues concern Codex/Work allowance visibility or attribution. This request covers the separate fixed-count regular Chat allowance, which OpenAI's documentation explicitly distinguishes from Work/Codex.

This is not a request to increase, reset, or change the quota. It is a request to expose the current state of an existing quota already enforced by the service.

If ownership of regular-Chat quota UI sits with another team or tracker, please route/transfer this request to that owner rather than closing it without a destination.

Start with Settings → Usage and the model picker requirements, then review related issues #24182, #29647, #35166, and #41617 to identify existing usage surfaces and ownership. Done means an authoritative server-provided used/remaining/reset view for regular Chat, shared-model labeling, and consistent presentation across web, desktop, and mobile.