Category

Code

AI that writes, tests, reviews and repairs software: research on coding agents and code models, with the benchmarks and limits that matter in practice.

Follow Code
IIIOpen sourcePreprint

Coding agents finish most tasks alone but only half of task chains

A new benchmark from Amazon researchers chains 20 coding requests together. Ten agents complete about three quarters of the single requests but only about half of the chains, mostly because they build on mistakes they never notice.