Why does GPT-6 Astra drain my Codex 5-hour limit so quickly?

The published estimates make the gap explicit. The help center's table of 'estimated local messages per five-hour period' gives GPT-6 Astra 5–45 on Plus, 25–225 on Pro 5x and 100–900 on Pro 20x, while GPT-5.6 Sol gets 10–100 / 50–500 / 200–2,000 and GPT-5.6 Luna 250–2,000 / 1,250–10,000 / 5,000–40,000 on the same plans. 'These are not fixed message limits. Actual usage varies by task, model and settings, and weekly limits may also apply.' On the token rate card, Astra is 250 credits per million input tokens, 25 cached and 1,250 output, against 100 / 10 / 500 for Sol and 5 / 0.5 / 30 for Luna.

Settings multiply the model cost. 'Different models can use different amounts of your allowance for the same task. Larger inputs and outputs, higher reasoning settings, Fast mode and tasks with multiple steps can also increase usage.' 'Higher effort can use more of your allowance and does not always produce a better result', and 'Fast mode applies a 2.5x multiplier to Astra's Standard rate.' The pricing page adds that 'larger projects, long-running tasks, or extended sessions that require the agent to hold more context will use significantly more per message.'

What to change, in the CLI: /model to 'Choose the active model (and reasoning effort, when available)'; /fast to toggle the Fast tier off; /compact to 'Summarize the visible chat to free tokens' or /new for a fresh chat; /status to read the '5h limit' bar before and after. OpenAI's advice on effort: 'Astra at Low effort can outperform Sol at High effort. If you've been getting good results with Sol at High, try Astra at Low or Medium as a starting point.' And a warning that switching alone is not a refill: 'Switching models does not restore allowance in a shared usage pool.' Issue #42987 (2026-09-05) is a user report of a Plus 5-hour allowance going 100% → 42% → 0% in two Astra Medium turns; treat it as one observation, not a rule.