Qwen Image Agent

Qwen Image Agent goes beyond simple image generation by introducing Planning, Reasoning, Searching, memory utilization, and feedback incorporation processes to fill in the user's missing context, thereby improving text-to-image generation performance.

This model was unveiled alongside a new benchmark, IA-Bench, to precisely measure agent-based image generation capabilities. IA-Bench tests the following key competencies.

• Planning: The ability to break down complex requests into executable steps
• Reasoning: The ability to compose images through logical thinking
• Search: The ability to retrieve necessary information externally
• Memory: The ability to remember and utilize prior context