Selection-Based Structured Reasoning: Toward Efficient Multimodal Search Agents

Apple researchers introduced Selection-based Structured Reasoning (SSR), a framework that reformulates reasoning as selection from pre-specified natural-language candidates rather than open-ended generation.

Evaluated on seven multimodal search benchmarks using 2B and 4B models, SSR achieved success rates competitive with leading search agents of the same scale while reducing per-turn reasoning latency by over 90% and total per-question model inference latency by 28–54%.