deepseek-v4.1POST
把 V4.1-Flash-1017 设为 Hermes 默认模型，将实测每任务成本
作者用 V4.1-Flash-1017 跑全天候的周期任务和研究，按每百万 token 输入 $0.15、输出 $0.60 的错峰价，打算统计实际 token 用量和单任务花费后回报。
Yannis
@yannisbuilds10 天前收录原推发布于 2026.10.17
Yannis @yannisbuilds原推发布于 2026.10.17
DeepSeek just released V4.1-Flash-1017. I’m making it my default in Hermes for recurring tasks and research. Those jobs run all day, so the cost adds up. Off-peak pricing is $0.15 input / $0.60 output per million tokens. The last version was pretty wordy, so I’ll check actual token usage and cost per task. I’ll report back once I’ve run it for a while.
这条是引用着另一条写的：x.com/deepseek_ai/status/2111375640717033777
x.com/yannisbuilds/status/2111390780799451538采集于 2026.10.18 16:44