deepseek-v4.1POST
Users report poor token efficiency with DeepSeek V4.1 Flash
A Japanese user notes widespread complaints about DeepSeek V4.1 Flash's token efficiency and is torn about sticking with GLM-5.3-Flash instead.
たまごとうみかち (みかち)
@micachi_netArchived 12 days agoOriginally posted 2026.10.19
たまごとうみかち (みかち) @micachi_netOriginally posted 2026.10.19
Deepseek V4.1 Flash、やはりトークン効率が悪いとあちこちで言われてるなあ GLM-5.3-Flash のトークン効率に不満がないだけに、やはり据え置いたほうがいいのか…？と悩み
x.com/micachi_net/status/2112011514383307226Archived 2026.10.19 12:01