SwiBench

Language Agents: From Reasoning to Acting — with Shunyu Yao of OpenAI, Harrison Chase of LangGraph

Latent Space

A benchmark for evaluating coding agents, designed by scraping GitHub and solving real-world engineering tasks.