Models
Available Ollama models
qwen2.5:32b
Light • Worker
Available
phi3:mini
Fast • Router
~2.2 GB • Fastest responses
mistral:7b-instruct-q4_0
General Purpose
~4.4 GB • Versatile
gemma2:9b
Capable • Heavy Worker
~5.4 GB • Best quality
gemma2:2b
Light • Fast
~1.6 GB • Lightweight
qwen2.5:3b
Light • Worker
~1.9 GB • Good balance
mistral:7b
General Purpose
~4.4 GB • Versatile
mxbai-embed-large:latest
Embeddings
~669 MB • Embeddings only
Recommended Roles
| Role | Recommended Model | Use Case |
|---|---|---|
| Router | phi3:mini | Fast classification, simple routing decisions |
| Light Worker | qwen2.5:3b | Quick parsing, template fills, simple tasks |
| Heavy Worker | gemma2:9b | Complex reasoning, content generation, analysis |
| Fallback | mistral:7b | General purpose backup, versatile |
| Memory | mxbai-embed-large | Embeddings for RAG and context retrieval |