Qwen2.5 14B Instruct
PermissiveAlibaba · Qwen · Released 2024-09
14BText
A sweet-spot size for local deployment: noticeably stronger reasoning than 7B while still quantizing down to a single consumer GPU.
Strengths
- +Apache 2.0 license
- +Good balance of quality vs local hardware requirements
- +128K context window
Limitations
- -Needs quantization to fit comfortably on 16GB consumer GPUs
- -Outclassed by Qwen2.5-32B on hard reasoning tasks
License
Apache 2.0
use commercially with attribution niceties
Hardware
runs on a good consumer GPU (or Apple Silicon) with quantization
Links
Stats
Qwen/Qwen2.5-14B-Instruct— downloads·— likes
via Hugging Face