Is it just me, or is GPT-6 Astra absolutely DEVOURING tokens?

Users are reporting that GPT-6 Astra consumes tokens at an "insanely high" rate compared to GPT-5.6 Sol, with even a few prompts quickly depleting usage meters. This high token burn is particularly noticeable in Codex and heavier reasoning modes, leading some to question Astra's practicality for regular users despite its power. The discussion highlights concerns about the intelligence-to-usage-cost ratio and whether the quality improvement over Sol justifies the increased consumption.

This report is among the first to detail user-reported token consumption rates for GPT-6 Astra, comparing them directly with GPT-5.6 Sol.