Primary sourceTogether AI
View sourceEvent ·Multi-source
NVIDIA：比价目表没用，token 效率才是选模型的关键指标
NVIDIA 的 Jacob Loewenstein 在 Open by Design NYC 活动上提出，只用价目表比较模型成本会漏掉一半的账，token 效率应排在模型评估标准前列。这一观点同时适用于两种场景：企业直接选用现成模型，以及团队自己训练模型。活动上还有 Reflection 的 Zach Wentz 和 Together AI 的 Hassan El Mghari，与在场开发者讨论了开源模型、post-training 和推理相关话题。
Current read
NVIDIA 的人在 Open by Design NYC 活动上说，选模型别只看单价，token 效率才是决定真实成本的关键，训练自己模型的团队也适用。现场还有 Reflection 和 Together AI 的人一起聊开源模型和 post-training，做技术选型的可以听听他们的算法。
6 sources34° AI heatLast updated 2026/11/9 21:09:22
Evidence
6 sourcesRelated source 1NVIDIA AI
View sourceRelated source 2Thomas Wolf
View sourceRelated source 3TestingCatalog
View sourceRelated source 4Blognone
View sourceRelated source 5elvis
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.