sphaela/Qwen3.6-35B-A3B-AutoRound-GGUF

Qwen3.6-35B-A3B GGUF (AutoRound Quantized, MTP Enabled)

This repository contains GGUF quantized versions of Qwen/Qwen3.6-35B-A3B created using Intel's AutoRound quantization method.

Qwen3.6-35B-A3B is a Mixture-of-Experts (MoE) model with 256 experts and approximately 3.6B active parameters.

🆕 MTP (Multi-Token Prediction) Support — All models now include the MTP / NextN head (blk.40.* tensors), enabling speculative decoding in compatible runtimes such as recent builds of llama.cpp. Each GGUF has been validated to contain the full set of MTP tensors.

Quantization Details

The models were quantized using various schemes provided by the auto-round tool with MTP layers explicitly enabled. For multimodal use, projector files (mmproj) are provided in F16, BF16, and F32 formats.

Files and Sizes

File Name	Quant Type	Size	Description
Qwen3.6-35B-A3B-Q2_K_S.gguf	Q2_K_S	12 GB	Extremely high compression, significant quality loss.
Qwen3.6-35B-A3B-Q2_K_MIXED.gguf	Q2_K_MIXED	13 GB	Recommended high-compression option. Fast inference.
Qwen3.6-35B-A3B-Q3_K_S.gguf	Q3_K_S	15 GB	Very high compression, notable quality loss.
Qwen3.6-35B-A3B-Q3_K_M.gguf	Q3_K_M	16 GB	Balanced 3-bit quantization.
Qwen3.6-35B-A3B-Q3_K_L.gguf	Q3_K_L	18 GB	High quality 3-bit quantization.
Qwen3.6-35B-A3B-Q4_0.gguf	Q4_0	19 GB	Standard 4-bit quantization, good balance.
Qwen3.6-35B-A3B-Q4_1.gguf	Q4_1	21 GB	Higher quality 4-bit quantization than Q4_0.
Qwen3.6-35B-A3B-Q4_K_S.gguf	Q4_K_S	19 GB	Small 4-bit K-quant, good efficiency.
Qwen3.6-35B-A3B-Q4_K_M.gguf	Q4_K_M	21 GB	Recommended 4-bit K-quant, excellent balance.
Qwen3.6-35B-A3B-Q5_0.gguf	Q5_0	23 GB	Standard 5-bit quantization, very high quality.
Qwen3.6-35B-A3B-Q5_1.gguf	Q5_1	25 GB	Higher quality 5-bit quantization than Q5_0.
Qwen3.6-35B-A3B-Q5_K_S.gguf	Q5_K_S	23 GB	Small 5-bit K-quant, very high quality.
Qwen3.6-35B-A3B-Q5_K_M.gguf	Q5_K_M	24 GB	Recommended 5-bit K-quant, near-lossless.
Qwen3.6-35B-A3B-Q6_K.gguf	Q6_K	28 GB	6-bit K-quant, virtually indistinguishable from F16.
Qwen3.6-35B-A3B-Q8_0.gguf	Q8_0	36 GB	8-bit quantization, near-lossless.
mmproj-model-f16.gguf	F16	—	Unified Projector in Float16 format.
mmproj-model-bf16.gguf	BF16	—	Unified Projector in BFloat16 format.
mmproj-model-f32.gguf	F32	—	Unified Projector in Float32 format.

Note: File sizes are slightly larger than non-MTP quants due to the additional MTP head weights.

Generate the Model

The models were generated using Intel's AutoRound with MTP layers explicitly enabled:

auto-round \
    --model Qwen/Qwen3.6-35B-A3B \
    --output_dir ./quantized/ \
    --scheme <SCHEME> \
    --iters 0 \
    --options '{"mtp_num_hidden_layers": 1, "num_nextn_predict_layers": 1}'

Usage with llama.cpp

These models can be used with a recent build of llama.cpp (must include Qwen3.5+ MTP support). For multimodal usage, specify the projector file:

./llama-cli -m Qwen3.6-35B-A3B-Q4_K_M.gguf --mmproj mmproj-model-f16.gguf --image your_image.jpg -p "Describe this image."

About AutoRound

AutoRound is an advanced quantization technique from Intel that aims to minimize accuracy loss through automated rounding optimization.

---

Support

These quantized models are made in my spare time using expensive hardware such as DGX Spark systems for quantization and validation. If you find these GGUFs useful for your projects, consider buying me a coffee to help cover hardware and compute costs. Every bit of support helps me keep producing high-quality quantized models for the community!

☕ Support me on Ko-fi