LLM EXPLORER 59,516 MODELS INDEXED

Qwen3.8 27B Brainwaves 2M Qx64 Hi Mlx by nightmedia

By nightmedia · 552 downloads
HF Repo ↗ Copy URL

Qwen3.8 27B Brainwaves 2M Qx64 Hi Mlx is an open-source language model by nightmedia. Features: 27b LLM, VRAM: 21GB, License: apache-2.0, Instruction-Based, LLM Explorer Score: 0.32.

  1m context   256k context   2m context   6-bit   All genres   All use cases Base model:adapter:davidau/qwe... Base model:davidau/qwen3.6-27b...   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   Qwen3.8   Qwen3 5   Qwen3 6   Qwen3 8   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-Brainwaves-2M-qx64-hi-mlx 

Qwen3.8 27B Brainwaves 2M Qx64 Hi Mlx Parameters and Internals

LLM NameQwen3.8 27B Brainwaves 2M Qx64 Hi Mlx
Repository 🤗https://huggingface.co/nightmedia/Qwen3.8-27B-Brainwaves-2M-qx64-hi-mlx 
Model Nameunsloth/Qwen3.8-27B
Base Model(s)  nbeerbower/Wichtel-Qwen3.6-27B   trohrbaugh/Qwen3.8-27B-heretic-ara   DavidAU/Qwen3.8-27B-Cold-Fusion-GAIN-V1.1   DavidAU/Qwen3.6-27B-V1.1-FF711-Darker-Hero-GAIN-H2.0   nightmedia/Qwen3.8-27B-Cold-Fusion-FF711-Darker-Hero-GAIN-B   nbeerbower/Wichtel-Qwen3.6-27B   trohrbaugh/Qwen3.8-27B-heretic-ara   DavidAU/Qwen3.8-27B-Cold-Fusion-GAIN-V1.1   DavidAU/Qwen3.6-27B-V1.1-FF711-Darker-Hero-GAIN-H2.0   nightmedia/Qwen3.8-27B-Cold-Fusion-FF711-Darker-Hero-GAIN-B
Model Size27b
Required VRAM21 GB
Updated2026-08-25
Maintainernightmedia
Model Typeqwen3_5
Instruction-BasedYes
Model Files  5.4 GB: 1-of-4   5.3 GB: 2-of-4   5.4 GB: 3-of-4   4.9 GB: 4-of-4
Supported Languagesen zh ja es
Model ArchitectureQwen3_5ForConditionalGeneration
Licenseapache-2.0
Model Max Length2097152
Transformers Version5.14.1
Tokenizer ClassTokenizersBackend
Padding Token<|endoftext|>
Errorsreplace

Best Alternatives to Qwen3.8 27B Brainwaves 2M Qx64 Hi Mlx

Best Alternatives
Context / RAM
Downloads
Likes
... 27B Brainwaves 1M Qx86 Hi Mlx0K / 27 GB4271
...1 Darker Hero GAIN B Mxfp8 Mlx0K / 28.7 GB6092
Tmax 27B0K / 53.8 GB463026
...sion GAIN V1.1 Fable Mxfp8 Mlx0K / 28.7 GB7202
Qwen3.8 27B Brainwaves0K / 55.5 GB4011
...en3.8 27B Brainwaves Mxfp8 Mlx0K / 28.7 GB3611
...ble Fusion F711 GAIN Mxfp4 Mlx0K / 15.1 GB4231
Qwen3.8 27b Euskara0K / 55.6 GB1820
Qwen27b Alia Tuned0K / 55.6 GB190
...3.5 27B Claude 4.6 OS INSTRUCT0K / 54.7 GB183713