Llama 3.1 8B

8B parameters
reasoningcoding
Parameters
8B
Context Length
128K
Architecture
llama
Release Date
2024

Available Quantizations

QuantizationFile SizeQuality
Q2_K3.2 GBLower
Q3_K_M4.0 GBLower
Q4_K_M4.9 GBGood
Q5_K_M5.6 GBGood
Q6_K6.4 GBHigh
Q8_08.5 GBHigh

Can it run? (Q4_K_M quantization)

GeForce RTX 4090
Runs great
~145 tok/s • 24GB VRAM
GeForce RTX 4070
Runs great
~73 tok/s • 12GB VRAM
GeForce RTX 3060 12GB
Runs great
~52 tok/s • 12GB VRAM
Radeon RX 7900 XTX
Runs great
~139 tok/s • 24GB VRAM
Check with your specific GPU →