Try Delta
← All models

Meta2024

Llama 3.1

Dense instruct

The 8B Llama 3.1 instruct checkpoint: still a reliable generalist for chat and writing when you want the Meta stack fully local.

  • Fits fromLaptop · 10 GB+
  • Contextup to 131K
  • Sizes1
  • RunsLocal on Delta
MultilingualGeneral chatWritingClassic Llama

Sizes

Llama 3.1 8B

4.9 GB • 131k ctx • Q4_K_M

Memory at context length

Approximate working set for Llama 3.1 8B as context grows.

4K
6.1 GB
8K
6.5 GB
16K
6.9 GB
33K
7.3 GB
66K
7.7 GB
131K
8.1 GB

Related families

Llama 3.2

Meta’s on-device Llama: lightweight 1B and 3B chat, with vision-class siblings.

Qwen 3.8

Qwen 3.8 Distill models: highly efficient and capable models optimized for edge and desktop.

Ministral 3

Mistral’s edge family with native vision: phones through workstations, fully local.