deepseek-v4.1POST
Skepticism over V4.1 Flash distilling GPT-6 Astra, citing token speed gap
The author pushes back on claims that V4.1-Flash-1017 distilled GPT-6 Astra, arguing DeepSeek's token speed is far too fast for that.
邦法
@bangbuildsArchived 12 days agoOriginally posted 2026.10.18
邦法 @bangbuildsOriginally posted 2026.10.18
都在说DeepSeek V4.1-Flash-1017 蒸馏了 GPT-6 Astra…… 讲道理，DeepSeek的这token速度，哪怕真蒸馏，连尾气都追不上吧🙈
x.com/bangbuilds/status/2111661677565575221Archived 2026.10.18 16:25