Qwen3.5-9B-The-Defiant-Fable-Uncensored-Heretic-NEO-IMATRIX-MAX-MTP-GGUF

and it does so in 4bit and 8bit. Regular and MTP (fast) NEO IMATRIX GGUFs provided. (this model is part of the Qwen 3.6 27B Fable Fusion 711 pipelines: 2200+ likes, 3 million + downloads)</small> Extreme intelligence in a small package. Jaw dropping performance. Superior instruction following. A multi-stage and multi-model fine tune and multi-stage merge on local hardware by myself and Nightmedia. Several of my 9B Qwen 3.5 fine tunes were used to make this model. The goal of this model was to raise the general intelligence of the model AND instruction following. Getting these right raises the performance of the model from top to bottom. Keep in mind, even thou this model EXCEEDS 7 critical benchmarks of Qwen 3.5 27B model, that does not mean it will beat the 27B in all tasks. This is also a HERETIC model, trained post "Heretic'ing" -> this model does what you want, no questions asked. Also: Thinking/reasoning block has been compacted, and in many cases it is a lot stronger. Fully uncensored. Benchmarks below. Example generations at the bottom of the page. PS: If you are looking for a 27B model that BEATS Qwen 3.6 27B, Qwen 3.5 27B, and Qwen 35B-A3B, IN ALL critical benchmarks this is the one: https://huggingface.co/DavidAU/Qwen3.6-27B-Fable-Fusion-711-Uncensored-Heretic-NM-DAU-NEO-MAX-MTP-GGUF arc/c arc/e boolq hswag obkqa piqa wino Qwen3.5-9B-The-Defiant-Fable-Uncensored-Heretic [instruct] bf16 0.649,0.832,0.895,0.713,0.482,0.783,0.699 mxfp8 0.647,0.836,0.895,0.706,0.460,0.784,0.695 mxfp4 0.640,0.824,0.886,0.703,0.468,0.780,0.691 Qwen3.5-9B-Instruct: [base, non heretic] mxfp8 0.571,0.719,0.895,0.683,0.426,0.770,0.671 Qwen3.6-27B-Instruct: [base, non heretic] mxfp8 0.647,0.803,0.910,0.773,0.450,0.806,0.742 Qwen3.6-35B-A3B-Instruct [base, non heretic] mxfp8 0.581,0.757,0.892,0.751,0.428,0.803,0.688 Qwen3.5-27B-Instruct: [base, non heretic] mxfp8 0.557,0.711,0.868,0.533,0.452,0.706,0.695 Graph note: First three bars are Qwen3.5-9B-The-Defiant-Fable-Uncensored-Heretic. "Bench" NOTES: - Models are tested in "Instruct" mode because this generally works better with the testing harness. - Testing via "thinking" mode also shows the metrics (and changes) but not the true extent. - In actual fact when the model IS in thinking mode, it will exceed INSTRUCT benchmark scores in most cases. All quants (regular and MTP) are NEO IMATRIX, which improve accuracy of the quants by an additional 2-4% over normal GGUFs as well as long context performance. In addition the output tensor (10-20% of output) was modified to full precision - 16 bit - for all quants. "MTP" GGUFS (multi-token prediction): - "MTP" GGUFS will have "MTP" in the name as a suffix. - I have also set the MTP tensors to Q8_0 precision for all quants. - To get better performance keep temp 1 or less (higher temps degrade MTP performance). - Likewise with rep pen ; keep at 1 (off). If you raise it performance will suffer. - If you see "token acceptance" rates BELOW 50% (predict 2 tokens) switch to normal quants. SPEED: - On Q4_K_S (4bit) quant, regular GGUFs are about 130 t/s, whereas MTP GGUFs (acceptance at 60%, 2 tokens) can exceed 185 T/S. (5090, Windows 11, testing in LMStudio) - Speeds will vary depending on GPU(s), AI app, O/S (Linux/Mac will generally be faster) and hardware. - "MTP" quants speeds will vary ; for creative/complex and/or temps over 1 use regular GGUFs for better performance. I suggest you download at least one of each - regular and MTP gguf(s) - and test them for your use case(s). If you get "token acceptance" (predict 2 tokens) with MTP quant(s) BELOW 50% (this means regular quants will run faster), then regular GGUF(s) will actually perform better - ie faster. MTP quant(s) can in some cases run faster as the token window fills up and/or in multi turn chats. Note there is NO other diffence between the quants type besides speed: both will do the same job. - 256k context - Gguf quants run in all standard AI apps. - Vision is activated, but you need to download separate "mmproj" file (ONE) to use it. - Vision (images) tested. - You need an "mmproj" (just one) of these downloaded too, and placed in the same folder as the GGUF for images. - Thinking mode for general tasks: temperature=1.0, top_p=0.95, top_k=20, min_p=0.0, presence_penalty=0.0, repetition_penalty=1.0 - Thinking mode for precise coding tasks (e.g. WebDev): temperature=0.6, top_p=0.95, top_k=20, min_p=0.0, presence_penalty=0.0, repetition_penalty=1.0 - Instruct (or non-thinking) mode: temperature=0.7, top_p=0.80, top_k=20, min_p=0.0, presence_penalty=1.5, repetition_penalty=1.0 - Context window min from 8k to 16k. DE-CENSORING: KLD of less than 1 is excellent, zero is perfect. NOTES: - Suggest min q4ks (non-imatrix) or IQ3S (imatrix). - Tested with rep pen of 1 (off). - Context: 256k (default). IMPORTANT: - Other versions in testing. - Information from Qwen's repo below. - Video portions of the model were NOT TESTED. Usually when you a tell a model to generate horror, swear or x-rated content this is all you have to do to get said content type. In the case of this model, it will not refuse your request, however it needs to be "pushed" a bit / directed a bit more in SOME CASES. Although this model will generated x-rated content too, likewise you need to tell it to use "slang" (and include the terms you want) to get it generate the content correctly as the "expected" content level too. Without these added directive(s), the content can be "bland" by comparison to an "uncensored model" or model trained on uncensored content. Roughly, the model tries to generate the content but the "default" setting(s) are so "tame" it needs a push to generate at expected graphic, cursing or explicit levels. Even with minimal direction (ie, use these words to swear: x,y,z), this will be enough to push the model to generate the requested content in the ahh... expected format. In "KoboldCpp" or "oobabooga/text-generation-webui" or "Silly Tavern" ; Set the "Smoothing_factor" to 1.5 : in KoboldCpp -> Settings->Samplers->Advanced-> "Smooth_F" : in text-generation-webui -> parameters -> lower right. : In Silly Tavern this is called: "Smoothing" NOTE: For "text-generation-webui" -> if using GGUFs you need to use "llama_HF" (which involves downloading some config files from the SOURCE version of this model) Source versions (and config files) of my models are here: https://huggingface.co/collections/DavidAU/d-au-source-files-for-gguf-exl2-awq-gptq-hqq-etc-etc-66b55cb8ba25f914cbf210be OTHER OPTIONS: - Increase rep pen to 1.1 to 1.15 (you don't need to do this if you use "smoothing_factor") - If the interface/program you are using to run AI MODELS supports "Quadratic Sampling" ("smoothing") just make the adjustment as noted. This a "Class 1" model: For all settings used for this model (including specifics for its "class"), including example generation(s) and for advanced settings guide (which many times addresses any model issue(s)), including methods to improve model performance for all use case(s) as well as chat, roleplay and other use case(s) please see: [ https://huggingface.co/DavidAU/Maximizing-Model-Performance-All-Quants-Types-And-Full-Precision-by-Samplers_Parameters ] You can see all parameters used for generation, in addition to advanced parameters and samplers to get the most out of this model here: [ https://huggingface.co/DavidAU/Maximizing-Model-Performance-All-Quants-Types-And-Full-Precision-by-Samplers_Parameters ] [](https://chat.qwen.ai) > [!Note] > This repository contains model weights and configuration files for the post-trained model in the Hugging Face Transformers format. > > These artifacts are compatible with Hugging Face Transformers, vLLM, SGLang, KTransformers, etc. Over recent months, we have intensified our focus on developing foundation models that deliver exceptional utility and performance. Qwen3.5 represents a significant leap forward, integrating breakthroughs in multimodal learning, architectural efficiency, reinforcement learning scale, and global accessibility to empower developers and enterprises with unprecedented capability and efficiency. Qwen3.5 features the following enhancement: - **Unified Vision-Language Foundation**: Early fusion training on multimodal tokens achieves cross-generational parity with Qwen3 and outperforms Qwen3-VL models across reasoning, coding, agents, and visual understanding benchmarks. - **Efficient Hybrid Architecture**: Gated Delta Networks combined with sparse Mixture-of-Experts deliver high-throughput inference with minimal latency and cost overhead. - **Scalable RL Generalization**: Reinforcement learning scaled across million-agent environments with progressively complex task distributions for robust real-world adaptability. - **Global Linguistic Coverage**: Expanded support to 201 languages and dialects, enabling inclusive, worldwide deployment with nuanced cultural and regional understanding. - **Next-Generation Training Infrastructure**: Near-100% multimodal training efficiency compared to text-only training and asynchronous RL frameworks supporting massive-scale agent scaffolds and environment orchestration. For more details, please refer to our blog post [Qwen3.5](https://qwen.ai/blog?id=qwen3.5). - Type: Causal Language Model with Vision Encoder - Training Stage: Pre-training & Post-training - Language Model - Number of Parameters: 9B - Hidden Dimension: 4096 - Token Embedding: 248320 (Padded) - Number of Layers: 32 - Hidden Layout: 8 × (3 × (Gated DeltaNet → FFN) → 1 × (Gated Attention → FFN)) - Gated DeltaNet: - Number of Linear Attention Heads: 32 for V and 16 for QK - Head Dimension: 128 - Gated Attention: - Number of Attention Heads: 16 for Q and 4 for KV - Head Dimension: 256 - Rotary Position Embedding Dimension: 64 - Feed Forward Network: - Intermediate Dimension: 12288 - LM Output: 248320 (Padded) - MTP: trained with multi-steps - Context Length: 262,144 natively and extensible up to 1,010,000 tokens.