Drew Breunig on Fable and the End of the Free Lunch in Coding Models

Drew Breunig reflects on a shift in artificial intelligence coding models following the release of a model called Fable. Collected on Simon Willison's weblog, the commentary highlights how developers previously felt little need to refine their coding harnesses or context strategies because cheaper and more capable new models routinely arrived to solve their problems automatically.

The release of Fable changed this dynamic because its operational cost was significantly higher than existing options, while other models like Opus remained sufficient for most routine coding tasks. This high cost forced developers to carefully evaluate which workloads required advanced capabilities and which tasks could be handled by cheaper alternatives.

This moment marks the end of an era where model upgrades consistently provided a free lunch by fixing inefficiencies without extra effort. Instead, users must now be more deliberate about how and where they deploy specific AI models to manage expenses effectively.