OpenAI adjusts GPT-6 Astra usage accounting, cutting subscription burn by as much as 75% in heavy workloads

OpenAI has optimized how ChatGPT subscription usage is counted for GPT-6 Astra, according to a BlockBeats flash update. Tibo, the company’s head of core product, said model quality remains unchanged. In some heavy-use scenarios, subscription quota consumption can now fall to roughly one-third to one-quarter of previous levels, implying a reduction of as much as 75%. The company did not disclose the exact mechanism behind the change. Based on the wording described in the report, the update appears more likely to involve subscription-side usage accounting rather than a change to the model itself. The adjustment was specifically framed as an optimization to ChatGPT subscription quotas. The move follows a wave of complaints over unusually fast quota depletion. Over the past two days, multiple reports appeared on OpenAI’s GitHub about Astra consuming limits too quickly. One Plus user said they hit the cap in about 20 minutes, while a Pro 20x user said a large portion of their weekly quota was used up within a single day.