Ollama (LLM)

Use Ollama as the LLM (dialogue + reasoning) provider.

Setup

Depending on your deployment, you may need an API key.

• Env var: OLLAMA_API_KEY

Example

from siphon.plugins import ollama

llm = ollama.LLM(
    model="llama3.1",
    base_url="http://localhost:11434/v1",
)

Common options

• model (default: llama3.1)
• base_url (default: http://localhost:11434/v1)
• api_key (default: None)
• temperature (default: 0.3)
• parallel_tool_calls (default: True)