Space Bunny（M3.1 Flash 匿名版）确认：Artificial Analysis 智能指数 42，OS
帖子梳理官方发布：Space Bunny 是 M3.1 Flash 早期版本，速度近三倍、成本降 70%、$0.15/M 输入，智能指数升 13 点到 42，OSWorld V2 中文榜登顶，10 月 23 日开源权重。
MiniMax just dropped M3.1 Flash and confirmed the mystery while doing it remember Space Bunny, the anonymous model that showed up on OpenRouter in late September with no company name attached? community testers linked it to MiniMax by probing its tokenizer today the chinese AI startup confirmed it. Space Bunny was an early build of M3.1 Flash the numbers: inference speed nearly tripled, costs down 70%, priced at $0.15 per million input tokens. that puts it directly against DeepSeek V4.1 Flash on price intelligence index jumped 13 points over M3 to 42 on artificial analysis. also topped the chinese leaderboard on OSWorld V2, which tests how well a model can use a computer on its own weights drop october 23 and a more capable Pro model is coming the chinese pricing race really does have no floor in sight. every few weeks someone undercuts the last one