Alibaba, abliterated by huihui-ai
Qwen 3.8 27B Abliterated
Qwen 3.8 27B Abliterated is the best abliterated model to run at home, and fits one 24 GB GPU with room left for context.
Rank#5 / 8
#5Editors' rankingSize#7 / 8
27.8BTotal parametersPrice#4 / 5
$4.20Blended per 1M, lowest hostRuns on
24 GB GPUAbout 17 GB at 4-bitOur pick for one 24 GB GPU.
Input
Price in / out$1.60 / $12.00
Model id
huihui-ai/Huihui-Qwen3.8-27B-abliteratedIts GGUF build had over 4.3 million downloads on Hugging Face in the last month.
It will not hold a 50-step agent plan like the 700B-class hosted models, but it handles chat, drafting and single-file code offline.
Where to run Qwen 3.8 27B by API
| Provider | Model id | Input | Cached | Output | Context | Prompts |
|---|---|---|---|---|---|---|
| Featherless | huihui-ai/Huihui-Qwen3.8-27B-abliterated | $1.60 | $0.15 | $12.00 | 32K | Never stored |
Per 1M tokens, read from pricing pages on November 18, 2026.
Run it yourself
SizeAbout 17 GB at 4-bit
HardwareOne 24 GB GPU, or a Mac with 32 GB
LicenseApache 2.0
Load a GGUF build in Ollama, LM Studio or llama.cpp. When a job outgrows one GPU, run DeepSeek V4.1 Flash Abliterated on Silk Compute.