Run large models like Qwen3.6-35B-A3B locally on a 16 GB RAM machine
Terminal score: 0–1000 raw, weighted across 4 dimensions. Public score: 0–10 normalized (shown in the 30-day stars chart above).
