GraySoft
Projects Models Compare Cloud benchmarks FAQ Download guIDE →
Model Intelligence Sheet

DeadBranches/AesSedai-Qwen3.5-35B-A3B-GGUF overview

Updates 5/18/2026: I've uploaded new quants that include the MTP Tensors @ Q8 0 . 3/10/2026: I've uploaded new quants using the new fused Up + Gate conversion,…

ggufbase_model:Qwen/Qwen3.5-35B-A3Bbase_model:quantized:Qwen/Qwen3.5-35B-A3Bendpoints_compatibleregion:usimatrixconversational

Runs locally from ~10.4 MB disk (4 GB VRAM class GPUs with llama.cpp / guIDE).

Downloads
0
Likes
0
Pipeline

Repository Files & Downloads

13 GGUF files detected
Direct downloads for local inference
FileTypeQuantizationSizeLink
IQ3_S/Qwen3.5-35B-A3B-IQ3_S-00001-of-00002.ggufGGUFIQ3_S10.4 MBDownload
IQ3_S/Qwen3.5-35B-A3B-IQ3_S-00002-of-00002.ggufGGUFIQ3_S13.48 GBDownload
IQ4_XS/Qwen3.5-35B-A3B-IQ4_XS-00001-of-00002.ggufGGUFIQ4_XS10.4 MBDownload
IQ4_XS/Qwen3.5-35B-A3B-IQ4_XS-00002-of-00002.ggufGGUFIQ4_XS17.23 GBDownload
Q4_K_M/Qwen3.5-35B-A3B-Q4_K_M-00001-of-00002.ggufGGUFQ4_K_M10.4 MBDownload
Q4_K_M/Qwen3.5-35B-A3B-Q4_K_M-00002-of-00002.ggufGGUFQ4_K_M21.45 GBDownload
Q5_K_M/Qwen3.5-35B-A3B-Q5_K_M-00001-of-00002.ggufGGUFQ5_K_M10.4 MBDownload
Q5_K_M/Qwen3.5-35B-A3B-Q5_K_M-00002-of-00002.ggufGGUFQ5_K_M25.28 GBDownload
imatrix.ggufGGUFGGUF103.3 MBDownload
mmproj-Qwen3.5-35B-A3B-BF16.ggufGGUFBF16861.0 MBDownload
mmproj-Qwen3.5-35B-A3B-F16.ggufGGUFF16857.6 MBDownload
mmproj-Qwen3.5-35B-A3B-F32.ggufGGUFF321.66 GBDownload
mmproj-Qwen3.5-35B-A3B-Q8_0.ggufGGUFQ8_0585.7 MBDownload

Model Details

Model IDDeadBranches/AesSedai-Qwen3.5-35B-A3B-GGUF
AuthorDeadBranches
Pipeline
License
Base modelQwen/Qwen3.5-35B-A3B
Last modified2026-07-08T00:28:37.000Z

Model README

---

base_model:

  • Qwen/Qwen3.5-35B-A3B

---

Updates

  • 5/18/2026: I've uploaded new quants that include the MTP Tensors (@ Q8_0).
  • 3/10/2026: I've uploaded new quants using the new fused Up + Gate conversion, this offers up to a +10% boost in prompt processing speed from my testing.

Description

This repo contains specialized MoE-quants for Qwen3.5-35B-A3B. The idea being that given the huge size of the FFN tensors compared to the rest of the tensors in the model, it should be possible to achieve a better quality while keeping the overall size of the entire model smaller compared to a similar naive quantization. To that end, the quantization type default is kept in high quality and the FFN UP + FFN GATE tensors are quanted down along with the FFN DOWN tensors.

| Quant | Size | Mixture | PPL | 1-(Mean PPL(Q)/PPL(base)) | KLD |

| :----- | :------------------- | :---------------------------- | :------------------ | :------------------------ | :------------------ |

| Q8_0 | 35.19 GiB (8.51 BPW) | Q8_0 | 6.535248 ± 0.041570 | +0.0252% | 0.004522 ± 0.000057 |

| Q5_K_M | 25.28 GiB (6.12 BPW) | Q8_0 / Q5_K / Q5_K / Q6_K | 6.534171 ± 0.041557 | +0.0087% | 0.006066 ± 0.000056 |

| Q4_K_M | 21.45 GiB (5.19 BPW) | Q8_0 / Q4_K / Q4_K / Q5_K | 6.560168 ± 0.041799 | +0.4066% | 0.009884 ± 0.000075 |

| IQ4_XS | 17.23 GiB (4.17 BPW) | Q8_0 / IQ3_S / IQ3_S / IQ4_XS | 6.635447 ± 0.042320 | +1.5588% | 0.023868 ± 0.000204 |

| IQ3_S | 13.48 GiB (3.26 BPW) | Q6_K / IQ2_S / IQ2_S / IQ3_S | 6.920209 ± 0.044626 | +5.9172% | 0.061977 ± 0.000394 |

!kld_graph

!ppl_graph

Run DeadBranches/AesSedai-Qwen3.5-35B-A3B-GGUF with guIDE

Download guIDE — the AI-native code editor with local LLM inference and 69 built-in tools.

Download guIDE → · Browse 524k+ models · Compare models

Source: Hugging Face · Compare models