Need real-time credit dashboard for Codex/ChatGPT

Codex
1 
Please add a real-time credit dashboard for ChatGPT/Codex.

Users need:
1. current credit balance
2. estimated credit cost before starting a task
3. live debit/usage while task is running
4. per-task credit history

Without this, users cannot control spending or trust credit usage. Please make credit/billing transparency clear and accessible.
2

Hi and welcome to the developer community.

Forget about the estimated credit cost. I don’t think that’s possible. OpenAI is far away from it. It is not even clear that a task will be solved by codex. How would it be possible to estimate something.

It also highly depends on how you use it and how clear your instructions are. Imagine you describe it badly and you get a result that was build exactly like you described but not even close to what youn meant?

But I agree on at least a token usage per turn should be displayed in realtime.

Or maybe also a budget per turn where you tell the bot to try with a budget of X and then stop if not successful within the budget.

3

Yes the estimates are random things. The only one I know that works (sort of)

  1. No-brainer
  2. Easy
  3. Can do
  4. Know how it can be done in theory
  5. Needs research
  6. Needs a deep study
  7. Sure

Where complexity of “no-brainer” is one and complexity of “sure” is infinity, the distribution of whatever is in the middle is literally “random whatever”

4

“But I agree on at least a token usage per turn should be displayed in realtime.”

when can you implement that please

5

This is a very valid UX and billing transparency request, especially for API-heavy workflows like Codex/ChatGPT automation.

From an implementation perspective, the key gap is the lack of a real-time usage metering layer exposed to users. Right now most systems expose:

  • aggregated billing data (delayed)
  • per-request token usage (post-response)

What’s missing is a live cost estimator + streaming usage ledger, ideally tied to:

  • pre-request token prediction (for cost estimation)
  • per-request incremental token tracking
  • session-level budget enforcement

A lightweight dashboard using event-streamed usage metrics (similar to Prometheus-style ingestion or OpenTelemetry traces) would solve this and also help prevent unexpected overages during long-running agent tasks.

6

So what’s your final answer?
Preview/Review kindly

7

Hey BAL, I agree that would be a valuable addition.

I'll pass this feedback along to the product team so they can evaluate whether to add this feature.

Avinash