今天新增deepseek-v4.1POST
退订 GPT Pro 转 Claude Max,子代理用 DeepSeek V4.1 Flash
作者因 OpenAI 10 月 30 日砍半额度而退订 GPT Pro,主模型换 Opus 5.5,子代理经 Baseten 调用 DeepSeek V4.1 Flash 1017,约 599 TPS 且极便宜。
Hayden
@0xHayd3n7 小时前收录原推发布于 2026.10.28
Hayden @0xHayd3n原推发布于 2026.10.28
Cancelling my GPT Pro plan at the end of the week. With OpenAI cutting the limits in half on October 30th, it’s hard to justify anymore, so I’m switching to Claude Max. Opus 5.5 will be my main orchestration model, and it scores higher on intelligence anyway. For subagents I’m using DeepSeek V4.1 Flash 1017 (fast) through @baseten. It’s ridiculously cheap and averages around 599 TPS, which makes it perfect for the grunt work.
x.com/0xhayd3n/status/2115250546928202088采集于 2026.10.28 12:01