Simon Willison's GPT-5.6 Notes: 'Sol on Medium' Is His New Coding Default, and the Model-Picker Is the Confusing Part

simonwillison.net·medium signal

In his July 9 annotated write-up, Willison flags Sol at Medium reasoning as a likely upgrade over his previous favorite (5.5 xhigh) for coding work, and highlights the new programmatic-tool-calling and multi-agent API additions as the genuinely interesting bits. He also notes GPT-5.6 Sol sets a new high of 53.6 on Agents' Last Exam, beating Claude Fable 5 by 13.1 points at roughly a quarter of the estimated cost — while calling out that figuring out which variant × reasoning-effort to use is now the most confusing part of adopting the family.