Vibe coded using local llm (24gb VRAM). Qwen family and Gemma.

Testing various backend (vllms, lm studio, ollama, llamacpp...) with various configuration (MoE layer offloading to CPU, concurrency, context size, KV cache sharing). Good fun

How to Play

Fill the grid so that every row, column, and 3x3 box contains the digits 1-9.