Downloads · 30 days
33
85% of all-time downloads
Parakon-27B
Parakon/Parakon-27B
Parakon-27B is a text generation model from Parakon. Use it when you need the model to write or continue text. The card lists the license as apache-2.0.
A compact, high-fidelity GGUF build for local inference, derived from Qwen/Qwen3.8-27B. It loads in any current llama.cpp build and the tools on top of it (LM Studio, Ollama, etc.) — no custom kernels.
All-time downloads
39
Public
Repo size
13.8 GB
Likes
1
Public
Repo makeup
Click a slice to open those files.
.gguf13.8 GB · 100%
At a glance
- Task
- Text Generation
- License
- apache-2.0
- Access
- Public
- Created
- Sep 2, 2026
- Updated
- Sep 2, 2026
- SHA
- 53b2aa60
Try a prompt
Base models
Model card
From the Hugging Face model README
Parakon-27B
A compact, high-fidelity GGUF build for local inference, derived from Qwen/Qwen3.8-27B. It loads in any current llama.cpp build and the tools on top of it (LM Studio, Ollama, etc.) — no custom kernels.
File
| File | Size | Quality vs Q8 (median KLD) | Perplexity |
|---|---|---|---|
Parakon-27B.gguf | 13.0 GB | 0.014 | −2.2% |
Measured against a full-precision Q8_0 reference: very low drift at a fraction of the
footprint. A median KLD under ~0.05 indicates minimal quality loss; above ~0.08 is real
degradation. These are relative-to-Q8 figures, not absolute task-benchmark scores.
Usage
llama-cli -m Parakon-27B.gguf -p "Your prompt here"
License
Derived from Qwen/Qwen3.8-27B and redistributed under its Apache-2.0 terms.
- Task
- Text Generation
- License
- apache-2.0
- Created
- Sep 2, 2026
- Updated
- Sep 2, 2026