Llama 3

by Meta · Website

Meta's Llama 3 is one of the most capable and widely adopted open-weight model families. Spanning from compact 1B parameter models to the massive 405B flagship, Llama 3 offers strong performance across reasoning, coding, and multilingual tasks. The series includes Llama 3.1, 3.2, and 3.3 releases with progressive improvements in efficiency and capability.

Variants (6)

Smallest: Llama 3.2 1B (1B)
Largest: Llama 3.1 405B (405B)

Llama 3.2 1B

Meta

1B
Min 2.1 GB
text-generation summarization

Llama 3.2 3B

Meta

3B
Min 3.3 GB
text-generation code-generation multilingual summarization

Llama 3.1 8B

Meta

8B
Min 6.3 GB
text-generation code-generation multilingual tool-use popular

Llama 3.1 70B

Meta

70B
Min 43.5 GB
text-generation code-generation reasoning multilingual

Llama 3.3 70B

Meta

70B
Min 43.5 GB
text-generation code-generation reasoning multilingual popular

Llama 3.1 405B

Meta

405B
Min 244.5 GB
text-generation code-generation reasoning multilingual