Qwen 3 4B

4.3B parameters
reasoningcodingmultilingual
Parameters
4.3B
Context Length
128K
Architecture
qwen2
Release Date
2025

Available Quantizations

QuantizationFile SizeQuality
Q2_K1.8 GBLower
Q3_K_M2.3 GBLower
Q4_K_M2.8 GBGood
Q5_K_M3.2 GBGood
Q6_K3.6 GBHigh
Q8_04.6 GBHigh

Can it run? (Q4_K_M quantization)

GeForce RTX 4090
Runs great
~271 tok/s • 24GB VRAM
GeForce RTX 4070
Runs great
~135 tok/s • 12GB VRAM
GeForce RTX 3060 12GB
Runs great
~97 tok/s • 12GB VRAM
Radeon RX 7900 XTX
Runs great
~258 tok/s • 24GB VRAM
Check with your specific GPU →