Models
Open models, optimized to run locally.
+ Add model
L
llama3 8Bvisiontools
Meta Llama 3 — most capable open model family
4.7 GB
M
mistral 7B
Mistral 7B — fast, capable, permissive
4.1 GB
φ
phi3 3.8B
Microsoft Phi-3 — small, sharp, mobile-grade
2.3 GB
Q
qwen2.5 7Bcode
Alibaba Qwen 2.5 — multilingual, strong reasoning
4.4 GB
G
gemma2 9B
Google Gemma 2 — research-grade open model
5.4 GB
D
deepseek-r1 7Bthinking
DeepSeek R1 — reasoning-first distillate
4.4 GB
N
nomic-embed-text 137Membedding
Nomic — high-performance text embeddings
274 MB
llama3
Meta's flagship open model. The Llama 3 family is the most capable open-weights model available, optimized for chat, reasoning, and instruction following.
Overview
Models
Tags
Versions
Parameters
8.0B
Quantization
Q4_0
Context
8,192
Size
4.7 GB
Architecture
llama
Updated
2 days ago
Run from the command line
~/projects — zsh
$ollama run llama3
>>> Tell me about the llama
Llamas are domesticated camelids native to South America...
>>>