DeepSWE benchmark reveals true differences between AI coding models

Original: This is the new standard for engineering evals https://t.co/UkEwUWybab

Source: x.com ↗