A note of thanks · Mac capability

Apple Foundation Models

Apple's on-device model gave us a useful floor to measure. Its limits on our action-grounding tasks clarified why the lab keeps its own weights and evaluation gate.

01 · The idea

What stayed with us

Apple's system model is a free and private routing floor, but its context and action-grounding limits make it unsuitable as the core capability dependency for a realistic tool catalog.

02 · The local translation

What we did with it

PostTrainLLM measured the framework, retained a bridge for comparison, ruled out adapter dependency, and kept owned weights plus owned evaluation as the differentiation.

03 · The boundary

Where the comparison stops

Core ML remains a possible deployment target for owned weights. The rejection applies to depending on Apple's closed model for capability, not to every Apple runtime technology.