Free download
The AI coding model cheat sheet.
Which model to use, when, and how hard to make it think. Claude Fable 5 and Opus 4.8 against GPT-5.6 Sol and Terra, task by task, on two printable pages. Current as of July 13, 2026.
Why the printed version
It is a reference, not a read. You do not want to scroll six thousand words at the moment you are picking a model. You want one page, open beside the terminal, that answers it in five seconds.
The whole table fits on one sheet. Ten tasks, each with the model, the effort level, and the reason. Page two carries every effort scale side by side, what the data actually says about
max, and the gotchas.It is the thing you hand to the team. Most of the cost of a bad model choice is other people quietly making it. Two pages travels; an article does not.
The three effort scales, side by side. Anthropic, the OpenAI API, and the Codex CLI each use a different set of levels and a different default. That comparison does not exist anywhere else, and it is the page people keep.
The advice itself is free. Every verdict, and the reasoning behind it, is published in full in the article, with every claim dated and every source linked. The cheat sheet is that advice in a format you can print. If you would rather just read it, do that instead. It costs nothing and it is the better version if you have twenty minutes.
Get the PDF
Two printable pages, current as of July 13, 2026. We will send it to the address below.
Next step
Want a second opinion on how your team is using these tools?
We help teams put real numbers on which model is actually worth it for their own codebase.