An Opinionated Guide to Which AI to Use for Tasks

In his blog post, Simon Willison reflects on the evolution of Ethan Mollick's guide to AI tools. A year ago, the guide focused on chat-based models like ChatGPT, Claude, and Gemini, with o3, Claude 4 Opus, and Gemini 2.5 Pro as the top models, and Deep Research as a useful alternative mode. Today, the emphasis is on agentic systems that can perform hours of human work in a single go. Gemini is no longer on the list because Google lacks an established entry in the Codex/ChatGPT Work/Cowork category, and Gemini Spark has yet to prove itself. Willison explains how to give ChatGPT or Claude access to your computer: in ChatGPT, the mode is called ChatGPT Work, while in Claude it's Cowork. The desktop apps also offer more advanced modes like Codex for ChatGPT and Code for Claude, but the naming is confusing. Willison notes that ChatGPT Work on mobile is essentially a version where the Code Interpreter container has internet access, which is unintuitive.