deepseek-v4.1POST
DeepSeek V4.1 Flash ranks second on a health Q&A benchmark at $0.005 per question
A user reports V4.1 Flash rose to second on a health Q&A leaderboard, beating GPT models and Opus 5.5 and trailing Sonnet 5.5 by just over a point, at $0.005 per question—about 1/16 of Sonnet's cost.
AI Wang
@loveAIwangArchived 8 days agoOriginally posted 2026.10.20
AI Wang @loveAIwangOriginally posted 2026.10.20
健康问答榜出来了，DeepSeek V4.1 Flash直接杀到了第二名！！！😱 GPT全家、Opus 5.5全被它压在下面了 距离第一的Sonnet 5.5就差一分多。 一道题才0.005刀，Sonnet得掏它16倍的钱😱😱😱 一个Flash小模型干成这样，我是真服了🤣 我已经开始琢磨这个月的API能省多少了
This post quotes another post: “Artificial Analysis launches HealthBench leaderboards, GPT-6 Astra on top”
x.com/loveaiwang/status/2112358984128839703Archived 2026.10.20 12:03