What is AesSedai/Qwen3.8-Flash-Next-GGUF?

--- base_model: - Qwen/Qwen3.8-Flash-Next --- ## Updates / Notes - 10/03/2026: I've re-quanted all of the models to include MTP, please re-download. - 09/02/2026: Added the PLEQ4_0 for IQ2_S and IQ3_S - 09/01/2026: The original quants all have Q8_0 for the engram embedding tensors, and I've uploaded three variants with Q4_0 for the embedded tensors. The only difference is the Q8_0 vs Q4_0 for those tensors. I'm leaving the original quants up for those who want to use the Q8_0 PLE's. This repo contains specialized MoE-quants for Qwen/Qwen3.8-Flash-Next. 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 U…

What license applies to AesSedai/Qwen3.8-Flash-Next-GGUF?

License: See model card. Verify terms on Hugging Face before commercial use.

How do I run AesSedai/Qwen3.8-Flash-Next-GGUF locally?

Download a GGUF file from this page and load it in guIDE or llama.cpp. Pipeline task: text-generation.

Model Intelligence Sheet

AesSedai/Qwen3.8-Flash-Next-GGUF overview

Updates / Notes 10/03/2026: I've re quanted all of the models to include MTP, please re download. 09/02/2026: Added the PLEQ4 0 for IQ2 S and IQ3 S 09/01/2026:…

ggufbase_model:Qwen/Qwen3.8-Flash-Nextbase_model:quantized:Qwen/Qwen3.8-Flash-Nextendpoints_compatibleregion:usimatrixconversational

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

Downloads

5,915

Likes

Pipeline

—

Author

AesSedai

Repository Files & Downloads

48 GGUF files detected

Direct downloads for local inference

File	Type	Quantization	Size	Link
IQ2_S-PLEQ4_0/Qwen3.8-Flash-Next-IQ2_S-PLEQ4_0-00001-of-00003.gguf	GGUF	IQ2_S	10.4 MB	Download
IQ2_S-PLEQ4_0/Qwen3.8-Flash-Next-IQ2_S-PLEQ4_0-00002-of-00003.gguf	GGUF	IQ2_S	46.41 GB	Download
IQ2_S-PLEQ4_0/Qwen3.8-Flash-Next-IQ2_S-PLEQ4_0-00003-of-00003.gguf	GGUF	IQ2_S	33.38 GB	Download
IQ2_S/Qwen3.8-Flash-Next-IQ2_S-00001-of-00005.gguf	GGUF	IQ2_S	10.4 MB	Download
IQ2_S/Qwen3.8-Flash-Next-IQ2_S-00002-of-00005.gguf	GGUF	IQ2_S	522.4 MB	Download
IQ2_S/Qwen3.8-Flash-Next-IQ2_S-00003-of-00005.gguf	GGUF	IQ2_S	50.66 GB	Download
IQ2_S/Qwen3.8-Flash-Next-IQ2_S-00004-of-00005.gguf	GGUF	IQ2_S	46.47 GB	Download
IQ2_S/Qwen3.8-Flash-Next-IQ2_S-00005-of-00005.gguf	GGUF	IQ2_S	5.99 GB	Download
IQ3_S-PLEQ4_0/Qwen3.8-Flash-Next-IQ3_S-PLEQ4_0-00001-of-00003.gguf	GGUF	IQ3_S	10.4 MB	Download
IQ3_S-PLEQ4_0/Qwen3.8-Flash-Next-IQ3_S-PLEQ4_0-00002-of-00003.gguf	GGUF	IQ3_S	46.20 GB	Download
IQ3_S-PLEQ4_0/Qwen3.8-Flash-Next-IQ3_S-PLEQ4_0-00003-of-00003.gguf	GGUF	IQ3_S	35.93 GB	Download
IQ3_S/Qwen3.8-Flash-Next-IQ3_S-00001-of-00005.gguf	GGUF	IQ3_S	10.4 MB	Download
IQ3_S/Qwen3.8-Flash-Next-IQ3_S-00002-of-00005.gguf	GGUF	IQ3_S	522.4 MB	Download
IQ3_S/Qwen3.8-Flash-Next-IQ3_S-00003-of-00005.gguf	GGUF	IQ3_S	50.66 GB	Download
IQ3_S/Qwen3.8-Flash-Next-IQ3_S-00004-of-00005.gguf	GGUF	IQ3_S	46.51 GB	Download
IQ3_S/Qwen3.8-Flash-Next-IQ3_S-00005-of-00005.gguf	GGUF	IQ3_S	8.29 GB	Download
IQ4_XS-PLEQ4_0/Qwen3.8-Flash-Next-IQ4_XS-PLEQ4_0-00001-of-00003.gguf	GGUF	IQ4_XS	10.4 MB	Download
IQ4_XS-PLEQ4_0/Qwen3.8-Flash-Next-IQ4_XS-PLEQ4_0-00002-of-00003.gguf	GGUF	IQ4_XS	46.27 GB	Download
IQ4_XS-PLEQ4_0/Qwen3.8-Flash-Next-IQ4_XS-PLEQ4_0-00003-of-00003.gguf	GGUF	IQ4_XS	44.88 GB	Download
IQ4_XS/Qwen3.8-Flash-Next-IQ4_XS-00001-of-00005.gguf	GGUF	IQ4_XS	10.4 MB	Download
IQ4_XS/Qwen3.8-Flash-Next-IQ4_XS-00002-of-00005.gguf	GGUF	IQ4_XS	669.2 MB	Download
IQ4_XS/Qwen3.8-Flash-Next-IQ4_XS-00003-of-00005.gguf	GGUF	IQ4_XS	50.66 GB	Download
IQ4_XS/Qwen3.8-Flash-Next-IQ4_XS-00004-of-00005.gguf	GGUF	IQ4_XS	46.46 GB	Download
IQ4_XS/Qwen3.8-Flash-Next-IQ4_XS-00005-of-00005.gguf	GGUF	IQ4_XS	17.21 GB	Download
Q4_K_M-PLEQ4_0/Qwen3.8-Flash-Next-Q4_K_M-PLEQ4_0-00001-of-00004.gguf	GGUF	Q4_K_M	10.4 MB	Download
Q4_K_M-PLEQ4_0/Qwen3.8-Flash-Next-Q4_K_M-PLEQ4_0-00002-of-00004.gguf	GGUF	Q4_K_M	46.55 GB	Download
Q4_K_M-PLEQ4_0/Qwen3.8-Flash-Next-Q4_K_M-PLEQ4_0-00003-of-00004.gguf	GGUF	Q4_K_M	46.14 GB	Download
Q4_K_M-PLEQ4_0/Qwen3.8-Flash-Next-Q4_K_M-PLEQ4_0-00004-of-00004.gguf	GGUF	Q4_K_M	15.45 GB	Download
Q4_K_M/Qwen3.8-Flash-Next-Q4_K_M-00001-of-00005.gguf	GGUF	Q4_K_M	10.4 MB	Download
Q4_K_M/Qwen3.8-Flash-Next-Q4_K_M-00002-of-00005.gguf	GGUF	Q4_K_M	669.2 MB	Download
Q4_K_M/Qwen3.8-Flash-Next-Q4_K_M-00003-of-00005.gguf	GGUF	Q4_K_M	50.66 GB	Download
Q4_K_M/Qwen3.8-Flash-Next-Q4_K_M-00004-of-00005.gguf	GGUF	Q4_K_M	46.45 GB	Download
Q4_K_M/Qwen3.8-Flash-Next-Q4_K_M-00005-of-00005.gguf	GGUF	Q4_K_M	34.21 GB	Download
Q5_K_M-PLEQ4_0/Qwen3.8-Flash-Next-Q5_K_M-PLEQ4_0-00001-of-00004.gguf	GGUF	Q5_K_M	10.4 MB	Download
Q5_K_M-PLEQ4_0/Qwen3.8-Flash-Next-Q5_K_M-PLEQ4_0-00002-of-00004.gguf	GGUF	Q5_K_M	46.56 GB	Download
Q5_K_M-PLEQ4_0/Qwen3.8-Flash-Next-Q5_K_M-PLEQ4_0-00003-of-00004.gguf	GGUF	Q5_K_M	46.11 GB	Download
Q5_K_M-PLEQ4_0/Qwen3.8-Flash-Next-Q5_K_M-PLEQ4_0-00004-of-00004.gguf	GGUF	Q5_K_M	36.56 GB	Download
Q5_K_M/Qwen3.8-Flash-Next-Q5_K_M-00001-of-00006.gguf	GGUF	Q5_K_M	10.4 MB	Download
Q5_K_M/Qwen3.8-Flash-Next-Q5_K_M-00002-of-00006.gguf	GGUF	Q5_K_M	669.2 MB	Download
Q5_K_M/Qwen3.8-Flash-Next-Q5_K_M-00003-of-00006.gguf	GGUF	Q5_K_M	50.66 GB	Download
Q5_K_M/Qwen3.8-Flash-Next-Q5_K_M-00004-of-00006.gguf	GGUF	Q5_K_M	45.88 GB	Download
Q5_K_M/Qwen3.8-Flash-Next-Q5_K_M-00005-of-00006.gguf	GGUF	Q5_K_M	46.50 GB	Download
Q5_K_M/Qwen3.8-Flash-Next-Q5_K_M-00006-of-00006.gguf	GGUF	Q5_K_M	9.38 GB	Download
imatrix.gguf	GGUF	GGUF	553.2 MB	Download
mmproj-Qwen3.8-Flash-Next-BF16.gguf	GGUF	BF16	865.5 MB	Download
mmproj-Qwen3.8-Flash-Next-F16.gguf	GGUF	F16	862.1 MB	Download
mmproj-Qwen3.8-Flash-Next-F32.gguf	GGUF	F32	1.67 GB	Download
mmproj-Qwen3.8-Flash-Next-Q8_0.gguf	GGUF	Q8_0	588.1 MB	Download

Model Details

Model ID	AesSedai/Qwen3.8-Flash-Next-GGUF
Author	AesSedai
Pipeline	—
License	—
Base model	Qwen/Qwen3.8-Flash-Next
Last modified	2026-10-03T10:36:56.000Z

Model README

---

base_model:

Qwen/Qwen3.8-Flash-Next

---

Updates / Notes

10/03/2026: I've re-quanted all of the models to include MTP, please re-download.
09/02/2026: Added the PLEQ4_0 for IQ2_S and IQ3_S
09/01/2026: The original quants all have Q8_0 for the engram embedding tensors, and I've uploaded three variants with Q4_0 for the embedded tensors. The only difference is the Q8_0 vs Q4_0 for those tensors. I'm leaving the original quants up for those who want to use the Q8_0 PLE's.

This repo contains specialized MoE-quants for Qwen/Qwen3.8-Flash-Next. 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.

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

| Q5_K_M | 153.07 GiB (7.32 BPW) | Q8_0 / Q5_K / Q5_K / Q6_K | 4.596526 ± 0.027937 | +0.1270% | 0.020218 ± 0.000145 |

| Q4_K_M | 131.98 GiB (6.31 BPW) | Q8_0 / Q4_K / Q4_K / Q5_K | 4.613527 ± 0.028075 | +0.4973% | 0.034206 ± 0.000237 |

| IQ4_XS | 114.99 GiB (5.50 BPW) | Q8_0 / IQ3_S / IQ3_S / IQ4_XS | 4.739310 ± 0.029099 | +3.2373% | 0.076158 ± 0.000486 |

| IQ3_S | 105.98 GiB (5.07 BPW) | Q6_K / IQ2_S / IQ2_S / IQ3_S | 4.863539 ± 0.030050 | +5.9434% | 0.160378 ± 0.000915 |

| IQ2_S | 103.63 GiB (4.96 BPW) | Q6_K / IQ2_XS / IQ2_XS / IQ3_XXS | 5.043883 ± 0.031497 | +9.8718% | 0.205353 ± 0.001121 |

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

| Q5_K_M (Q4_0 PLE) | 129.23 GiB (6.18 BPW) | Q8_0 / Q5_K / Q5_K / Q6_K | 4.620130 ± 0.028097 | +0.6412% | 0.030943 ± 0.000212 |

| Q4_K_M (Q4_0 PLE) | 108.14 GiB (5.17 BPW) | Q8_0 / Q4_K / Q4_K / Q5_K | 4.629426 ± 0.028160 | +0.8437% | 0.043475 ± 0.000297 |

| IQ4_XS (Q4_0 PLE) | 91.14 GiB (4.36 BPW) | Q8_0 / IQ3_S / IQ3_S / IQ4_XS | 4.759109 ± 0.029230 | +3.6686% | 0.084406 ± 0.000531 |

| IQ3_S (Q4_0 PLE) | 82.13 GiB (3.93 BPW) | Q6_K / IQ2_S / IQ2_S / IQ3_S | 4.889509 ± 0.030205 | +6.5091% | 0.167198 ± 0.000940 |

| IQ2_S (Q4_0 PLE) | 79.79 GiB (3.82 BPW) | Q6_K / IQ2_XS / IQ2_XS / IQ3_XXS | 5.062449 ± 0.031623 | +10.2763% | 0.211550 ± 0.001142 |

!kld_graph

!ppl_graph

Run AesSedai/Qwen3.8-Flash-Next-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