AWS Certified AI Practitioner AIF-C01 Practice Question

Because the model will run locally on a resource-constrained mobile device and must respond in real time, the amount of compute and memory the model consumes is the dominant concern. Smaller, less complex models generally load faster and perform inference with lower latency on limited hardware. While factors such as supported input length, language coverage, or licensing cost may influence the overall choice, they do not have as direct an impact on meeting strict on-device latency targets as the model's size and resource requirements.