LLM EXPLORER 62,696 MODELS INDEXED

Qwen3.8 27B Cold Fusion GAIN V1.1 Fable Mxfp8 Mlx by nightmedia

By nightmedia · 720 downloads
HF Repo ↗ Copy URL

Qwen3.8 27B Cold Fusion GAIN V1.1 Fable Mxfp8 Mlx is an open-source language model by nightmedia. Features: 27b LLM, VRAM: 28.7GB, License: apache-2.0, Instruction-Based, LLM Explorer Score: 0.28.

  1m context   256k context   8-bit   All genres   All use cases Base model:adapter:davidau/qwe... Base model:davidau/qwen3.8-27b...   Bf16   Chain-of-thought   Claude-distillation   Claude4.6   Coding   Conversational   Creative   Creative writing   Deckard(qx)   Distillation   En   Endpoints compatible   Es   Experimental   Fable   Fiction   Fiction story   Fiction writing   Image-text-to-text   Instruct   Instruction-tuned   Ja   Long-cot   Lora   Math   Merge   Mergekit   Mlx   Multilingual   Plot generation   Polaris   Polaris-alpha   Qwen   Qwen3.5   Qwen3.6   Qwen3 5   Qwen3 6   Reasoning   Region:us   Research   Roleplaying   Safetensors   Scene continue   Science fiction   Sft   Sharded   Stem   Story   Story generation   Storytelling   Sub-plot generation   Tensorflow   Unsloth   Vivid prosing   Vivid writing   Writing   Zh
Model Card on HF 🤗: https://huggingface.co/nightmedia/Qwen3.8-27B-Cold-Fusion-GAIN-V1.1-Fable-mxfp8-mlx 

Qwen3.8 27B Cold Fusion GAIN V1.1 Fable Mxfp8 Mlx Parameters and Internals

LLM NameQwen3.8 27B Cold Fusion GAIN V1.1 Fable Mxfp8 Mlx
Repository 🤗https://huggingface.co/nightmedia/Qwen3.8-27B-Cold-Fusion-GAIN-V1.1-Fable-mxfp8-mlx 
Model Nameunsloth/Qwen3.8-27B
Base Model(s)  DavidAU/Qwen3.8-27B-Cold-Fusion-GAIN-V1.1   armand0e/Qwen3.8-27B-Fable-Distill   DavidAU/Qwen3.8-27B-Cold-Fusion-GAIN-V1.1   armand0e/Qwen3.8-27B-Fable-Distill
Model Size27b
Required VRAM28.7 GB
Updated2026-08-17
Maintainernightmedia
Model Typeqwen3_5
Instruction-BasedYes
Model Files  5.3 GB: 1-of-6   5.3 GB: 2-of-6   5.3 GB: 3-of-6   5.3 GB: 4-of-6   5.3 GB: 5-of-6   2.2 GB: 6-of-6
Supported Languagesen zh ja es
Model ArchitectureQwen3_5ForConditionalGeneration
Licenseapache-2.0
Model Max Length262144
Transformers Version5.14.1
Tokenizer ClassTokenizersBackend
Padding Token<|endoftext|>
Errorsreplace

Best Alternatives to Qwen3.8 27B Cold Fusion GAIN V1.1 Fable Mxfp8 Mlx

Best Alternatives
Context / RAM
Downloads
Likes
...27B USS Collective Qx64 Hi Mlx0K / 21 GB4172
Qwopus3.8 27B Flash V20K / 55.6 GB12617
...en3.8 27B Parapraxis Mxfp8 Mlx0K / 28.7 GB2061
...en3.8 27B MindMeld Qx64 Hi Mlx0K / 21 GB3461
Qwopus3.8 27B Flash0K / 55.6 GB23419
Tmax 27B0K / 53.8 GB463026
Qwen3.8 27B FULL NVFP40K / 18.8 GB3362
...8 27B Brainwaves WFH Mxfp4 Mlx0K / 15.1 GB4901
...1 Darker Hero GAIN B Mxfp8 Mlx0K / 28.7 GB6092
... 27B Brainwaves 2M Qx64 Hi Mlx0K / 21 GB5521
Note: green Score (e.g. "73.2") means that the model is better than nightmedia/Qwen3.8-27B-Cold-Fusion-GAIN-V1.1-Fable-mxfp8-mlx.