back

Qwen 3.8 27B

Apache 2.0

Alibaba · 27B · Dense

Native multimodal dense Qwen 3.8 model for coding, office automation, and agentic workflows

HuggingFace
0 downloads 0 likes 2026-08 256K context

Use Cases

chat vision reasoning code

Quantization Options

Quant Bits VRAM Quality Status
Q2_K 2 9.1 GB low —
Q3_K_M 3 12.6 GB moderate —
Q4_K_M 4 14.3 GB good —
Q5_K_M 5 17.8 GB good —
Q6_K 6 21.2 GB excellent —
Q8_0 8 28.2 GB excellent —
F16 16 55.8 GB lossless —

About this model

HF model card Qwen
Qwen 3.8 27B source artwork

Alibaba

27B

Dense model for chat, vision, reasoning workloads.

Context

256K

Q4 VRAM

~14.3 GB

Image saved from public model sources

Dense Thinking Tool use Vision Code

Qwen 3.8 27B on your hardware

Native multimodal dense Qwen 3.8 model for coding, office automation, and agentic workflows. This page turns the Hugging Face model card into practical local-run numbers, so you can compare quantized VRAM, system RAM, and expected fit before downloading a large checkpoint.

Model shape

27B total parameters. Dense models use the whole network for each token.

Local fit target

Start with Q4_K_M: about 13.2 GB on disk and 14.3 GB VRAM before extra context and runtime overhead.

Best use cases

chat, vision, reasoning, code

License and source

Listed as Apache 2.0 from huggingface.co/Qwen/Qwen3.8-27B.

Want the real verdict? Pick your GPU or edit the specs on this page and compare the quant table below.

Open HF repo

Benchmark snapshot

Public eval numbers from the model card or benchmark indexes. Scores use each benchmark's own scale, so compare rows by task type, not as one combined rating.

Source
8 public scores Qwen/Qwen3.8-27B Hugging Face model card

Coding

LiveCodeBench v6

90.3

Reasoning

GPQA Diamond

89.2

Agentic multimodal

OSWorld-Verified

84.3

Instruction following

IFBench

79.5

Long-horizon work

CoWorkBench

70.7

Browser use

WebArena-Verified

64.8

Document vision

OmniDocBench 1.5

91.1

Knowledge

HLE

30.8