What are the hardware requirements for Qwen3.8 27B Kestrel?

Qwen3.8 27B Kestrel requires approximately 21.9 GB of VRAM. Quantized variants may run on less VRAM; see the Quantized Models section on this page.

Who developed Qwen3.8 27B Kestrel and how large is it?

Qwen3.8 27B Kestrel is developed by glyd, 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 27B Kestrel?

Qwen3.8 27B Kestrel is hosted on Hugging Face and linked from this page. LLM Explorer also lists quantized variants and similar alternatives if available.

Qwen3.8 27B Kestrel by glyd — VRAM 21.9GB

Name: Qwen3.8 27B Kestrel
Author: glyd

Qwen3.8 27B Kestrel is an open-source language model by glyd. Features: 27b LLM, VRAM: 21.9GB, License: apache-2.0, LLM Explorer Score: 0.29.

8-bit Base model:quantized:qwen/qwen... Base model:qwen/qwen3.8-27b Conversational Eval-results Glyd Kestrel Lossy Qwen3 5 Region:us Safetensors Sharded Tensorflow

Model Card on HF 🤗: https://huggingface.co/glyd/Qwen3.8-27B-kestrel

Qwen3.8 27B Kestrel Parameters and Internals

LLM Name	Qwen3.8 27B Kestrel
Repository 🤗	https://huggingface.co/glyd/Qwen3.8-27B-kestrel
Base Model(s)	Qwen/Qwen3.8-27B Qwen/Qwen3.8-27B
Model Size	27b
Required VRAM	21.9 GB
Updated	2026-10-13
Maintainer	glyd
Model Type	qwen3_5
Model Files	1.0 GB: 1-of-22 1.0 GB: 2-of-22 1.0 GB: 3-of-22 1.1 GB: 4-of-22 1.0 GB: 5-of-22 1.0 GB: 6-of-22 1.1 GB: 7-of-22 1.0 GB: 8-of-22 1.0 GB: 9-of-22 1.1 GB: 10-of-22 1.0 GB: 11-of-22 1.0 GB: 12-of-22 1.1 GB: 13-of-22 1.0 GB: 14-of-22 1.0 GB: 15-of-22 1.1 GB: 16-of-22 1.0 GB: 17-of-22 1.0 GB: 18-of-22 1.1 GB: 19-of-22 1.0 GB: 20-of-22 1.2 GB: 21-of-22 0.1 GB: 22-of-22
Model Architecture	Qwen3_5ForConditionalGeneration
License	apache-2.0
Model Max Length	262144
Transformers Version	5.8.0.dev0
Tokenizer Class	Qwen2Tokenizer
Padding Token	<\|endoftext\|>
Errors	replace

Rank the Qwen3.8 27B Kestrel 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 27B Kestrel

Best Alternatives	Context / RAM	Downloads	Likes
...wopus3.8 27B Flash Mlx OQ4e 1M	1024K / 16.9 GB	211	1
Agens Pilot	1024K / 54.7 GB	82	3
Qwen3.8 27B NPU2	256K / GB	252	4
JEV 27B VL	0K / 55.6 GB	1525286	966
Clef	0K / 54.7 GB	15064	2103
Qwen3.6 27B	0K / 55.4 GB	6728138	2197
Qwen3.8 27B	0K / 55.6 GB	2	8753
Qwen3.8 27B	0K / 55.6 GB	2727	21
Qwen3.8 27B MLX	0K / 16 GB	1411	18
Qwen3.8 27B FP8	0K / 0.4 GB	123157	416

Note: green Score (e.g. "73.2") means that the model is better than glyd/Qwen3.8-27B-kestrel.