Llama 3.1 8B

balanced

Great balance of speed and quality. Recommended starting point.

Parameters8.0B
Context131,072 tokens
Download4.7 GB
Min RAM8 GB
Architecturellama

Available quantizations

QuantSizeQuality
Q4_K_M4.7 GBgood
Q5_K_M5.5 GBbetter
Q8_08.1 GBbest

Check whether your machine can run this model — with real measured speeds from anonymous community benchmarks — on the Central-Intel compatibility page.