Space Bunny (anonymous M3.1 Flash) hands-on: two hard prompts, self-tested and self-fixed
The author ran two hard prompts against anonymous Space Bunny in OpenCode: it built a Connect Four engine with alpha-beta pruning and a monitoring dashboard, running its own tests and fixing bugs, with the official release stronger on agentic work.
last month i tested space bunny in opencode with two hard prompts it built a connect four engine with minimax, alpha-beta pruning and a live telemetry panel, then a full monitoring dashboard from an ascii wireframe. it wrote the files, ran its own tests and fixed its own bugs before handing them over that model is now revealed as MiniMax M3.1 Flash, and the official release is stronger on exactly this kind of agentic work - 62.8 on Automation-Bench, end-to-end workflows - 76.2 on BankerToolBench, multi-step planning and chained tool calls - 73.7% on DeepSWE v1.1, fixing and validating code through tests if space bunny impressed you, try the official version at http://agent.minimax.io
This post quotes another post: x.com/sakatayasha/status/2102969148934689207