同一 prompt 实测：M3.1 Flash 40 分钟免费，DeepSeek V4.1 Flash 50 分钟 $0
作者用同一 prompt、最高推理档跑两个模型：MiniMax M3.1 Flash 耗时 40 分钟、成本 $0，DeepSeek V4.1 Flash 耗时 50 分钟、成本 $0.4；两者风格不同难分胜负，旧版 M3 做不出这种效果。
MiniMax M3.1 Flash vs DeepSeek v4.1-1017 Flash tested both models with same prompt at highest reasoning available and results came out really different > minimax took 40 mins and cost $0 > deepseek took 50 mins and cost $0.4 really surprised by how well minimax did. old m3 couldn't make anything like this also both models have their own style and it's hard for me to pick winner here so you decide. which one did better?
i am running tests via minimax plan in mcode and model is free there that's why cost zero it's also free on openrouter so run some tests if you want. it's a good model for daily use https://openrouter.ai/minimax/minimax-m3.1-flash:free