Give Pro users a consistent credit rate and an auditable usage ledger

ChatGPT Feature requests
1

I pay for Pro and use it for ongoing projects. I want fully auditable usage accounting and consistent treatment of subscription allowances and purchased credits. A percentage remaining does not tell me enough to budget for finishing a project or another day of work. I would be more willing to buy additional credits if I could reasonably estimate what they would get me.

My proposal:

  1. Use the same credit unit for subscription allowances and purchased credits. State how many credits each plan includes and when they reset. If purchased credits are discounted, offer more credits per dollar.

  2. Publish one rate card. The same model, reasoning setting, speed mode, and billable usage should consume the same number of credits regardless of where those credits came from. Models can have different rates, but the conversion should be clear. Publish rate changes in advance and retain the rate used on each transaction.

For speed modes, my preferred pricing is Standard at 1×, Fast at 2×, and Ultra Fast at 6×, applied consistently to included and purchased credits. These are proposed rates, not a claim about today’s pricing. For equivalent billable usage with the same model and settings, the credit charge should be the same regardless of whether those credits came with the subscription or were purchased separately. That would remove the concern that using a speed mode on a subscription allowance carries an extra penalty. A charge multiplier should also be distinguished from a speed claim such as “up to 8× faster.”

  1. Provide an exportable usage ledger by project and task. Show the model, reasoning and speed settings, billable input/output and cached tokens where applicable, tool or other charges, credits deducted, and the rate used. Show retries and failed or interrupted work and explain whether they were charged. Make it possible to reconcile the ledger with the remaining balance.

  2. Add a budgeting calculator based on actual usage. Show typical task costs and a forecast range, with clear assumptions. Work varies, so estimates will not be guarantees. Let users set project spending caps and receive warnings before reaching them.

For a hypothetical example, if comparable work averages $68 per day in additional-credit costs and I need four more days, I can budget about $272. I could buy enough to finish, reduce the workload, or fund only the most important stage.

Without that information, I hesitate to buy $100 of credits because I don’t know whether that buys an hour or several days of my normal work. Predictability would make purchasing easier and let users audit the charges instead of guessing whether changes in the percentage meter are fair.

I would also like usage resets made available again, with clear pricing and terms.

Could OpenAI provide a consistent published rate card, an itemized usage export, and a calculator that uses the same accounting for included and purchased credits?