The chart on this link compares all qwen3.5 models down to 0.8B.

https://www.reddit.com/r/LocalLLaMA/comments/1ro7xve/qwen35_...