Based on Qwen3.5-122B-A10B and converted from the MLX Community 4-bit release (https://huggingface.co/mlx-community/Qwen3.5-122B-A10B-4bit)
419 Pulls 1 Tag Updated 2 months ago
Experimental - Qwen 3.6, MTP-enabled, 512k context (20GB KV cache footprint with OLLAMA_KV_CACHE_TYPE=q8_0)
1,288 Pulls 4 Tags Updated 2 months ago
Quants from Q2 up to Q5 from Unsloth K_M and UD_K_XL and builds for 12, 16 and 24 GB
12.6K Pulls 22 Tags Updated 1 week ago