Qwen 2.5 14B

14.8B parameters
reasoningcodingmultilingual
Parameters
14.8B
Context Length
128K
Architecture
qwen2
Release Date
2024

Available Quantizations

QuantizationFile SizeQuality
Q2_K5.5 GBLower
Q3_K_M7.0 GBLower
Q4_K_M8.6 GBGood
Q5_K_M9.8 GBGood

Can it run? (Q4_K_M quantization)

GeForce RTX 4090
Runs great
~79 tok/s • 24GB VRAM
GeForce RTX 4070
Runs great
~39 tok/s • 12GB VRAM
GeForce RTX 3060 12GB
Runs great
~28 tok/s • 12GB VRAM
Radeon RX 7900 XTX
Runs great
~75 tok/s • 24GB VRAM
Check with your specific GPU →