What Apple’s on-device model can and can’t do: 30 real tasks tested.
中文摘要
通过30项实际任务测试，评估苹果端侧AI模型的性能边界，探讨其能力与局限性。
English Summary
An evaluation of Apple's on-device AI model through 30 real tasks, highlighting its capabilities and limitations.
Original Excerpt
Naive prompts, tool calling, and full app relaunches — 30 tasks, testing where Apple’s on-device model holds up and where it quietly breaks. Continue reading on Medium »