minimax-m3.1POST
实测吐槽：M3.1 Flash 与 DeepSeek V4.1 Flash 都过度思考
用户同时用过 MiniMax M3.1 Flash 和 DeepSeek V4.1 Flash，认为两者都很强但思考 token 浪费严重，希望提升 token 效率。
KC
@karanC_128 天前收录原推发布于 2026.10.19
KC @karanC_12原推发布于 2026.10.19
MiniMax M3.1 Flash & DeepSeek V4.1 Flash are both crazy good, but they overthink way too much. Maybe that's the price of getting this much intelligence so cheap. Imagine what they could do with better token efficiency.
这条是引用着另一条写的：「M3.1 Flash 一次生成宝塔，作者称优于 DeepSeek V4.1 Flash」
x.com/karanc_12/status/2112193980695982393采集于 2026.10.20 20:16