Primary sourcearXiv: OpenAI
View sourceEvent ·Official first-hand
谁为 KV Cache 买单?unalloc 拆解共享推理成本归属
论文提出开源工具 unalloc,把 OpenCost、LiteLLM 以及 OpenAI、Anthropic 的成本数据合并成一本精确台账,并标出无人认领的支出占比。在构造的多 pod 部署场景中,只给 LeaderWorkerSet 的 leader pod 打 owner 标签会让该部署 66% 的 GPU 账单无主,默认回退键则把 61% 的支出归到一个 Helm chart 名下,未分配比例表面降至 4%。数据源口径同样有坑:同时开启所有来源会把网关支出全部重复计算,只读一页账单 API 只能看到约四分之一的支出。在 NVIDIA H100 上跑 vLLM 时,按 token 计量比按时间份额计量让检索密集型租户多摊 12-14 个百分点的账单,而各负载下 GPU 利用率读数都停在 97-99%。论文还把这组结果与近期的 Shapley 能耗归因研究做了对照。
Current read
unalloc 把 Kubernetes 和 LLM API 账单拼成一本账,还量化出按 token 和按时间计费能差 12 个点。
11 sources36° AI heatLast updated 2026/9/21 17:57:40
Evidence
11 sourcesRelated source 1AI Engineer
View sourceRelated source 2VTV Công nghệ
View sourceRelated source 3Blognone
View sourceRelated source 4shao__meng
View sourceRelated source 5向阳乔木
View sourceRelated source 6宝玉
View sourceRelated source 7a16z
View sourceRelated source 8Decoder
View sourceRelated source 9Tech in Asia
View sourceRelated source 10IT之家
View sourceConflict check
Dedup data only shows that these sources discuss the same event, not that they agree. There is no structured supporting or refuting evidence yet, so no conflict is inferred automatically.