← Back to Models

qwen3.5-35b-a3b

mlx-community/qwen3.5-35b-a3b

View on HuggingFace
Publisher
mlx-community
Architecture
qwen3_5_moe(sparse)
Parameters
35B
Quantization
4bit (4.0 bpw)
Format
mlx
Max Context
262k tokens
Type
llm
File Size
19.02 GB
Est. RAM
20.5 GB
Est. VRAM
20.5 GB(~3B active)
Capabilities
visiontrained_for_tool_use

Benchmark Results

Compare Models