NVIDIA 论文：频繁压缩历史的智能体几乎不损失准确率
NVIDIA 实测发现让智能体频繁压缩历史几乎不掉分，Claude Opus 4.5 压缩后 SWE-bench 还高了 0.2 分，能省好几倍成本。
NVIDIA 的论文测试了让智能体定期压缩上下文的做法：到达设定上限时写一份交接笔记，只带任务和笔记重启，长文档改用 Bash 工具按需检索。成本差异显著，按 256K 而非 16K 窗口服务开源模型，每 token 生成成本高 3.6 到 7 倍。在 6 个长程基准上，每个模型 22 种压缩设置中最多 5 种比完整历史多丢 2 分以上。Claude Opus 4.5 每步都压缩时在 SWE-bench Verified 拿到 70.2%，比完整历史的 70.0% 还高。
New NVIDIA paper finds that AI agents barely lose accuracy when they keep contexts short by compacting their history often.
Each new token costs more to generate as the context grows. Serving open models at a 256K context window instead of 16K cost 3.6x to 7x more per generated token.
They had agents stop at a set limit, write a short handoff note, and restart with only the task and that note. For long documents, the agent searched files with Bash tools instead of reading everything at once.
On 6 long-horizon benchmarks, at most 5 of 22 compaction settings per model lost more than 2 points versus full history. Compacting after every action kept Claude Opus 4.5 at 70.2% on SWE-bench Verified, versus 70.0% with full history.
Before paying for long context, give agents a structured handoff note and file search tools, then check if short contexts match your accuracy.
