What are the hardware requirements for Qwen3.8 P44w75 16.8B Unrepaired?

Qwen3.8 P44w75 16.8B Unrepaired requires approximately 2.4 GB of VRAM and supports a context window of 256K tokens. Quantized variants may run on less VRAM; see the Quantized Models section on this page.

Who developed Qwen3.8 P44w75 16.8B Unrepaired and how large is it?

Qwen3.8 P44w75 16.8B Unrepaired is developed by logic65, a model with 27b parameters. The model is published as open weights on Hugging Face and indexed on LLM Explorer with full benchmark history.

Where can I download or evaluate Qwen3.8 P44w75 16.8B Unrepaired?

Qwen3.8 P44w75 16.8B Unrepaired is hosted on Hugging Face and linked from this page. LLM Explorer also lists quantized variants and similar alternatives if available.

Qwen3.8 P44w75 16.8B Unrepaired by logic65 — VRAM 2.4GB, 256K context

Name: Qwen3.8 P44w75 16.8B Unrepaired
Author: logic65

Qwen3.8 P44w75 16.8B Unrepaired is an open-source language model by logic65. Features: 27b LLM, VRAM: 2.4GB, Context: 256K, License: apache-2.0, LLM Explorer Score: 0.28.

Base model:quantized:qwen/qwen... Base model:qwen/qwen3.8-27b-fp... Conversational Depth-pruning Endpoints compatible Gated-deltanet Gguf Pruning Qwen3 5 Region:us Safetensors Width-pruning Zero-training

Model Card on HF 🤗: https://huggingface.co/logic65/Qwen3.8-p44w75-16.8B-unrepaired

Qwen3.8 P44w75 16.8B Unrepaired Parameters and Internals

LLM Name	Qwen3.8 P44w75 16.8B Unrepaired
Repository 🤗	https://huggingface.co/logic65/Qwen3.8-p44w75-16.8B-unrepaired
Base Model(s)	Qwen/Qwen3.8-27B-FP8 Qwen/Qwen3.8-27B-FP8
Model Size	27b
Required VRAM	2.4 GB
Updated	2026-08-16
Maintainer	logic65
Model Type	qwen3_5
Model Files	2.5 GB 4.0 GB 4.0 GB 3.9 GB 4.0 GB 3.9 GB 4.0 GB 4.0 GB 2.4 GB
Model Architecture	Qwen3_5ForCausalLM
License	apache-2.0
Context Length	262144
Model Max Length	262144
Tokenizer Class	Qwen2Tokenizer
Padding Token	<\|endoftext\|>
Vocabulary Size	248320
Torch Data Type	bfloat16
Errors	replace

Rank the Qwen3.8 P44w75 16.8B Unrepaired capabilities

Have you tried this model? Rate its performance — this feedback helps the ML community find the right model for their needs.

Instruction Following and Task Automation
Factuality and Completeness of Knowledge
Censorship and Alignment
Data Analysis and Insight Generation
Text Generation
Text Summarization and Feature Extraction
Code Generation
Multi-Language Support and Translation

Best Alternatives to Qwen3.8 P44w75 16.8B Unrepaired

Best Alternatives	Context / RAM	Downloads	Likes
Hemmingway 1	256K / 53.8 GB	11181	846
JEV 27B	256K / 53.9 GB	178643	79
OrcaSAQ 2 27B	256K / 12.3 GB	1330	169
Qwen3.6 27B OBLITERATED	256K / 53.6 GB	15137	184
Darwin 27B RSI	256K / 53.8 GB	459	14
Mnma Qwen3.8 27b Nvfp4	256K / 18.8 GB	527	11
Qwen3.8 27B Code Tools Merged	256K / 53.8 GB	559	3
Sharona Q27B R CodeSecurity V2	256K / 53.8 GB	674	1
Qwen3.8 27B Obliterated E03	256K / 53.6 GB	708	1
...2 Abliterated OptiQ 3.7bpw Mlx	256K / 13.4 GB	6727	14

Note: green Score (e.g. "73.2") means that the model is better than logic65/Qwen3.8-p44w75-16.8B-unrepaired.