[Codex app] GPT-6 Astra Ultra consumes ~30% of a 20x Pro weekly quota during a ~1-hour task #43029

What version of the Codex App are you using (From “About Codex” dialog)?

26.901.31953 — released September 3, 2026.

The About dialog identifies the application as ChatGPT — Powered by Codex & OWL; the affected workflow is in the app's Codex interface.

What subscription do you have?

ChatGPT Pro — 20x usage tier.

What platform is your computer?

• macOS Tahoe 26.6.2
• MacBook Pro, 16-inch, Apple M5 Max
• 128 GB memory

What issue are you seeing?

GPT-6 Astra Ultra appears to consume an excessive amount of the weekly Codex allowance, even on the 20x Pro tier.

During a roughly one-hour coding task, approximately 30% of my weekly allowance was consumed. This is the weekly quota, not merely a short-term usage window. The task view shows approximately 1h 11m of work.

The task was a SwiftUI loading-indicator redesign in an existing app, involving implementation, compilation, and testing. The selected model is displayed as GPT-6 Astra Ultra, and the session shows 3 completed subagents.

I appreciate the model's capabilities, but spending roughly a third of the weekly allowance during one task feels disproportionate for a 20x Pro subscription and makes sustained development difficult.

The 30% figure is my observed allowance consumption during this task. Exact per-run token counts and server-side quota accounting are not included in this report, so I cannot determine whether this reflects intended model weighting, agent-related overhead, an accounting problem, or other account activity. Please use the submitted diagnostics to verify the attribution and calculation.

What steps can reproduce the bug?

1. Use the Codex interface in the macOS app with a ChatGPT Pro 20x subscription.
2. Select GPT-6 Astra Ultra.
3. Note the remaining weekly allowance, then run a coding task involving planning, implementation, compilation, and testing. In the reported session, the task ran for approximately one hour and included subagents.
4. Compare the weekly allowance afterward. In this case, approximately 30% of the weekly allowance was consumed during that period.

These steps describe the affected workflow; consistent reproduction and session-specific attribution need to be checked against the usage records.

What is the expected behavior?

Usage should be predictable and reasonably sustainable for ongoing development on a 20x Pro subscription. I understand that a more capable model and agent activity may consume more resources, but the current observed drain feels excessive.

Please investigate whether the deduction is correct. If this is intended behavior, please clearly explain Astra Ultra's quota weighting, make per-task consumption easier to understand, and review whether the weekly allowance is adequate for practical use at this subscription tier.

Additional information

Feedback ID: 01a07010-3842-7a70-ae9a-7f1f1a0cf370

The in-app feedback submission succeeded, with diagnostic logs enabled.

Please investigate:

• How much of the observed weekly quota reduction is attributable to this session, including its subagents, versus any other account activity.
• Whether model weighting, context/caching accounting, retries, or duplicated metering contributed to the drain. These are investigation questions, not confirmed causes.
• Whether the account's 20x Pro entitlement was applied correctly.
• If the deduction is intentional, whether quota transparency and the available allowance can be improved.

Screenshots documenting the selected model, task duration, feedback confirmation, and app version are available but are not attached to this issue. The device serial number and unrelated project details have been omitted from this report.