gpt-6POST
质疑 Astra 效率排名：Claude 额度反而更耐用
作者吐槽基准说 GPT-6 Astra 是最高效前沿模型，但自己狂用 Opus 5.5 额度却像早期 Codex 一样宽裕。
Dwayne
@CtrlAltDwayne28 天前收录原推发布于 2026.10.06
Dwayne @CtrlAltDwayne原推发布于 2026.10.06
The benchmarks are claiming GPT-6 Astra is one of the most efficient frontier models right now. Then why is it I can't burn my Claude usage with Opus 5.5 if it's less efficient? I've been hammering Claude and it's like the early days of Codex, the usage limits are generous.
x.com/ctrlaltdwayne/status/2107412096141172959采集于 2026.10.06 21:24