我是担心降智，一直这么20w用着，倒是压缩很频繁
吗德 原来订阅不支持 还好没需要
Context Window Specifications
- API (gpt-5.5 & gpt-5.5-pro): 1,000,000 (1M) tokens.
- Codex (VS Code Extension/IDE): 400,000 (400K) tokens, though some reports indicate an effective usable window often closer to 258K–272K due to “compaction” and output headroom.
- Output Token Limit: 128,000 tokens
14 个帖子 - 9 位参与者
来源: linux.do查看原文