重跑 Space Bunny 旧测试:M3.1 Flash 已达 GLM-5.3-Flash 水平
开发者用同样测试对比匿名版与正式版,称差异巨大、旧问题消失,水平与 GLM-5.3-Flash 相当并优于初版 DeepSeek V4.1 Flash。
So @MiniMax_AI MiniMax-M3.1-Flash is not just Space-bunny-alpha, it's an upgrade! I tested Space-bunny a lot when it was available for free as a stealth model, including the example in this quoted post. After re-running the same tests with M3.1-Flash, i can tell you the difference is huge! I can still see some similarities, but it feels like a completely different version now. Almost everything got better, and those annoying issues i noticed with bunny aren't showing up anymore. It's crazy how much they improved in what, 3 weeks! From my testing, it's now on the same level as GLM-5.3-Flash and better than the first DeepSeek-V4.1-Flash. I've noticed some really good improvements. I'll be posting comparisons soon so you can see the difference yourselves. Now i'm really looking forward to how crazy the Pro model is going to be!
这条是引用着另一条写的:x.com/itsmeajaykv/status/2105291332017086853