1
DeepSeek V4还能更省！新工具缓存命中率高达99.82%，2折稳定到手
DeepSeek V4新工具缓存命中率99.82%，长会话成本直降80%，2折玩转大模型。
原本4亿+token、61美元的账单，直降至12美元
2
Stop Running LLM Workloads on Vanilla Kubernetes
K8s调度LLM虽顺手，但隔离不足的默认配置暗藏工具调用与数据泄露风险，云原生社区必须重新审视信任边界。
TL;DR: Kubernetes schedules LLM workloads well, but it does not give them the isolation boundary they need once they start calling tools, executing co…
3
If you are redlining the LLM, you aren't headlining
LLM的上下文窗口也有“红色区域”——接近极限时输出质量骤降，工具调用会异常。
Article URL: https://ghuntley.com/redlining/ Comments URL: https://news.ycombinator.com/item?id=48150288 Points: 2 # Comments: 1
4
The Oats Protocol – Open Agent Tools for Local Coding Agents
本地编码代理的新开源协议，让AI工具调用更透明可控，值得开发者关注
Recently I was using functiongemma and watched it load and run local source code as a tool call without any training/tuning. A couple days later I got…