AI CURATED HOT STORY · 10/19/2026
MiniMax has officially revealed that the anonymous model "Space Bunny," which topped leaderboards on OpenRouter and OpenCode, is its new M3.1 Flash. The model achieves nearly 3x faster inference speeds and a 70% reduction in overall usage costs while maintaining top-tier capabilities in Agentic tasks, coding, and Computer Use.
Cross-verified data indicates M3.1 Flash ranks 12th among open-source models on the Code Arena WebDev leaderboard, with a total score increase of 106 points over its predecessor M3. During its anonymous phase, it processed 56.7 trillion tokens, setting a record for single anonymous models, and showed significant improvements across multiple vertical domains.
This release marks a major breakthrough for lightweight, high-performance models in complex task execution. With weights scheduled to be open-sourced on October 23 and limited-time free trials available, MiniMax aims to lower barriers for local deployment and secondary development, reshaping cost-efficiency standards for high-frequency API calls.
The line compares only sources observed throughout; gap hours are interpolated to keep the trend continuous.