DeepSWE adds GPT-5.6 models to benchmark

DeepSWE has added the GPT-5.6 models to its benchmark for evaluating AI coding agents. A performance chart was shared on Reddit.