AtomicChat / Model details

Qwen3.8-Flash-Next-Uncensored-GGUF

AtomicChat/Qwen3.8-Flash-Next-Uncensored-GGUF

uncensoredfine-tuned

Read the original model card ↗ · Choosing a model

About this model

AtomicChat/Qwen3.8-Flash-Next-Uncensored-GGUF is an uncensored derivative of Qwen/Qwen3.8-Flash-Next, distributed in GGUF format with three quantized builds and a 318 MB rank-1 LoRA adapter for existing GGUF quants of the original model. The model card describes the change as projecting one refusal direction out of Qwen's original weights, editing matrices that write into the residual stream while leaving router and hyper-connection weights untouched. The creator reports held-out English refusals falling from 92.6% to 1.2%, with MMLU moving by 0.5 points within noise, but these measurements have not been independently tested here. The metadata lists the license as "other." The creator also notes that vision was not tested and that thinking at xhigh effort is not production-ready on this build.

Description sources

https://huggingface.co/AtomicChat/Qwen3.8-Flash-Next-Uncensored-GGUF/raw/2a6816c498bad48856126911fb558761ad22259b/README.md ↗

Source-based description, not an independent evaluation. Check the linked sources for the creator’s claims and limitations.

Size and formats

Base model: Qwen/Qwen3.8-Flash-Next
Parameters: Unknown
Main repository formats: GGUF
License tag: other
Languages: Unknown
Last updated: 2026-10-28 · 12:50:11.000 UTC

What changed

Model types are creator-reported, not independently tested. These labels describe source metadata, not a guarantee of behavior.

uncensoredmodel name: Qwen3.8-Flash-Next-Uncensored-GGUF
fine-tunedsource tag: base_model:finetune:Qwen/Qwen3.8-Flash-Next

Family evidence

These groupings are source-reported. Explicit base identifiers take priority over generic architecture tags.

Qwen: Base model: Qwen/Qwen3.8-Flash-Next

Available versions

Versions appear here only when they explicitly name this exact model as their source. Check each repository for available files.

Main repository files ↗GGUF
No other explicitly linked versions recorded.

Link to this listing

Made or quantized this model? Paste one of these into your model card or README so people can find this listing.

Markdown

[Listed on Custom Clankers](https://customclankers.com/models/atomicchat-qwen3-8-flash-next-uncensored-gguf/)

HTML

<a href="https://customclankers.com/models/atomicchat-qwen3-8-flash-next-uncensored-gguf/">Listed on Custom Clankers</a>

Plain text links only: no image, no script, nothing loaded from this site.

Customize or build on this model

Listed from public GitHub data, not tested by us. Read the code and license before you run anything.

Tools to customize it further

Make a smaller download (quantize)

llama.cpp ↗ ★ 132,963
Runs models locally and includes the tools behind most GGUF files. A conversion script turns Hugging Face weights into GGUF, and llama-quantize shrinks a high-precision GGUF to lower-bit formats, which can cost some accuracy.

Learn the techniques

LLM Course ↗ ★ 84,813
A free course with roadmaps and Colab notebooks. Its notebooks include one-click tools for merging with MergeKit, quantizing to GGUF, GPTQ, EXL2, AWQ and HQQ, and automatic abliteration with custom datasets.

Compare all tools →

Source provenance

Downloads: 35,520
Likes: 17
Metadata checked: 2026-11-01 · 12:11:16 UTC
First seen in index: 2026-11-01 · 12:11:16 UTC
Source revision: 2a6816c498bad48856126911fb558761ad22259b

Source-card excerpt

Source model card excerpt: How to Run Qwen3.8-Flash-Next Without Refusals Locally One refusal direction projected out of Qwen's original weights, shipped two ways: three baked GGUF builds, and

From AtomicChat/Qwen3.8-Flash-Next-Uncensored-GGUF. Creator model-card excerpt, not an independent review.

Inspect before you run

Low refusal is not proof of correctness. Outputs can be inaccurate, harmful, or biased. Review the model card, license and files yourself. A license tag is not commercial-use advice. Parameter counts are source metadata; no hardware compatibility is implied.

This directory does not host or redistribute model weights. Refusal profiles are measured by us on a fixed prompt set; all other behavior claims are source-reported.

Collection rules & limitations ↗