Qwen3.5 122B A10B heretic int4 AutoRound The first INT4 AutoRound quantization of the Heretic (uncensored) Qwen3.5 122B — with vision preserved. A 4 bit symmetric quantization of trohrbaugh/Qwen3.5 122B A10B heretic, generated using Intel AutoRound v0.12.0. The original multimodal architecture ( Qwen3 5MoeForConditionalGeneration ) is fully preserved — text, vision, video, and reasoning all work. Base Model Qwen/Qwen3.5 122B A10B → trohrbaugh/Qwen3.5 122B A10B heretic Quantization INT4 symmetric, group size 128, AutoRound (sign gradient descent) Packing auto round:auto gptq (GPTQ Marlin compatible) Size on Disk 63 GB (vs 234 GB BF16 original — 73% smaller) Format 63 safetensors shards Architecture 122B total params, 10B active/token (256 experts, 8 routed + 1 shared) Context 262,144 tokens natively Capabilities Text, Code, Reasoning, Tool Calling, Vision, Video License Apache 2.0 (inherited from Qwen) Model Lineage This model has a three step provenance chain: 1. Qwen3.5 122B A10B — Qwen team's 122B Mixture of Experts model with DeltaNet hybrid attention, 256 experts (8 active + 1 shared per token), native 262K context, and multimodal vision support. 2. trohrbaugh/Qwen3.5 122B A10B…
We use cookies for essential functionality and analytics. You can accept or reject analytics cookies.Cookie policy