deepseek-v4.1IMAGE

DeepSeek V4.1 Flash latency varies by channel: Cline Pass hits 47s to first token

Comparison finds official-channel first-token latency around 1.5s versus Cline Pass starting at 2.5s and reaching 47s at 335K context, while output speed stays near 265 tok/s.

向箔 Foil

@heyfoilArchived 10 days agoOriginally posted 2026.10.18

DeepSeek V4.1 Flash latency varies by channel: Cline Pass hits 47s to first token (image 1 of 2)

DeepSeek V4.1 Flash latency varies by channel: Cline Pass hits 47s to first token (image 2 of 2)

向箔 Foil @heyfoilOriginally posted 2026.10.18

Open original on X

一个很容易被忽略的大坑！同样是 DeepSeek V4.1 Flash 1017，换个调用渠道，速度能差十万八千里！！！有人实测官方渠道首 Token 延迟稳定在 1.5 秒左右但换成 Cline Pass 后，首 Token 延迟从 2.5 秒起步，随着上下文增长到 335K，甚至一度飙到了 47 秒！！😭😭😭 离谱的是，后续 Token 输出速度却基本一致，稳定在 265 tok/s 左右......

x.com/heyfoil/status/2111704609110831240Archived 2026.10.18 16:31