Large Language Model (LLM)-based agentic systems have shown strong capabilities across various tasks. However, existing multi-agent frameworks often rely on static or task-level workflows, which either over-process simple queries or underperform on complex ones, while also neglecting the efficiency-performance trade-offs across heterogeneous LLMs. To address these limitations, we propose Difficulty-Aware Agentic Orchestration (DAAO), which can dynamically generate query-specific multi-agent workflows guided by predicted query difficulty. DAAO comprises three interdependent modules: a variational autoencoder (VAE) for difficulty estimation, a modular operator allocator, and a cost- and performance-aware LLM router. A self-adjusting policy updates difficulty estimates based on workflow success, enabling simpler workflows for easy queries and more complex strategies for harder ones. Experiments on six benchmarks demonstrate that DAAO surpasses prior multi-agent systems in both accuracy and inference efficiency, validating its effectiveness for adaptive, difficulty-aware reasoning. Our code is open-sourced at https://github.com/AutoAgents-ai/DAAO

Person Re-Identification (ReID) is a critical task in computer vision that involves identifying individuals across different cameras or video frames. It’s challenging due to variations in appearance, lighting, viewpoints, clothing, and occlusions. Text-based ReID adds complexity by requiring image retrieval or individual identification from a dataset based on text queries. The BLIP-2 model addresses these challenges by combining multi-modal alignment and matching into a single framework, using a pre-trained vision-language model with a Q-Former component to bridge the visual and textual modalities. This approach significantly boosts performance in multi-modal tasks and information retrieval, especially with large datasets. The LV-ReID framework, which incorporates BLIP-2, enhances text-based ReID by integrating retrieval and generation tasks. The experimental results show BLIP-2’s effectiveness in aligning and matching pedestrian images for information retrieval tasks. It demonstrates proficiency in multi-modal tasks and offers an efficient solution for text-based ReID by fusing visual and textual data, improving pedestrian identification accuracy in complex environments.