agent = CodeAgent(tools=[], model=model) # no need for tools
agent.run("Describe these documents:", images=[document_1, document_2, document_3])