GraphicBench: A Planning Benchmark for Graphic Design with Language Agents

TL;DR

GraphicBench provides a challenging dataset of 1,079 graphic-design queries across four design types to probe LLM-driven planning for creative tasks. The authors introduce GraphicTown, an LLM agent framework with three design experts and 46 executable actions to generate, integrate, and execute design workflows in web environments. Experiments across six LLMs show that while agents can form plans that respect explicit and implicit constraints, real-world execution suffers from spatial reasoning gaps, cross-expert coordination issues, and action-retrieval errors, highlighting the gap between planning and execution in creative design. The work establishes GraphicBench as a valuable benchmark for advancing planning and execution in multimodal, creative domains and identifies concrete directions for improving multi-agent coordination and action retrieval in design tasks.

Abstract

Large Language Model (LLM)-powered agents have unlocked new possibilities for automating human tasks. While prior work has focused on well-defined tasks with specified goals, the capabilities of agents in creative design tasks with open-ended goals remain underexplored. We introduce GraphicBench, a new planning benchmark for graphic design that covers 1,079 user queries and input images across four design types. We further present GraphicTown, an LLM agent framework with three design experts and 46 actions (tools) to choose from for executing each step of the planned workflows in web environments. Experiments with six LLMs demonstrate their ability to generate workflows that integrate both explicit design constraints from user queries and implicit commonsense constraints. However, these workflows often do not lead to successful execution outcomes, primarily due to challenges in: (1) reasoning about spatial relationships, (2) coordinating global dependencies across experts, and (3) retrieving the most appropriate action per step. We envision GraphicBench as a challenging yet valuable testbed for advancing LLM-agent planning and execution in creative design tasks.

Figures (2)

• Figure 2: Workflow evaluation results for each model. We normalize color, text, and images pass rates to [0,1].
• Figure 3: Error distribution per model. Most errors arise from models failing to resolve dependencies or retrieving invalid actions.