14 upvotes
New Benchmark Tests AI Agents on Real Multimodal Coding Requests
What is Co-Design Bench: Multimodal Collaborative Coding Evaluation for Agentic Models?
Co-Design Bench offers a reference-free benchmark of 273 tasks sourced from 37,463 real user queries. It measures whether coding agents can turn multimodal requests into working application revisions without needing pre-written reference code.
Who it is for
Researchers and developers of agentic coding models can use it to gauge real-world co-design performance and guide improvements.