X.PIN· @thexpin · X·· 6 天前AI 评分67
AI 导读
MiniMax 发布了 M3.1 Flash 模型，其成本降至 M3 的三分之一，运行速度提升约 3 倍。该模型在 DeepSWE v1.1 上达到了 73.7%，速度超过 200 tokens/s，此前曾以 Space Bunny 名称在 OpenRouter 和 OpenCode 上匿名运行，计划于 10 月 23 日开源权重。
正文 · 原文
Chinese AI startup MiniMax released M3.1 Flash, which it says costs one-third of M3 and runs about 3x faster. It reports DeepSWE v1.1 at 73.7% and 200+ tokens/s. It plans to open the weights October 23. It ran anonymously as 'Space Bunny' on OpenRouter and OpenCode.
来源：X.PIN · x.com