Qwen/Qwen3.5-35B-A3B
Compact Qwen3.5 multimodal MoE (35B total / 3B active) with gated delta networks, 256 experts, and 262K context
Compact Qwen3.5 MoE — single-GPU FP8 or 2x GPU BF16 serving
Compact Qwen3.5 multimodal MoE (35B total / 3B active) with gated delta networks, 256 experts, and 262K context
Compact Qwen3.5 MoE — single-GPU FP8 or 2x GPU BF16 serving
View on HuggingFace
View on ModelScope