For the complete documentation index, see llms.txt. This page is also available as Markdown.
Tool/Function Calling
Function Calling
Ollama supports tool (function) calling via the /v1/chat/completions endpoint. For models like Llama 3.1 and Mistral Nemo, you can define tools in the request, and the model may return a tool_calls object with arguments to execute—just like OpenAI's function-calling.