Terminal-Bench 2.0 launches alongside Harbor, a new framework for testing agents in containers

The developers of Terminal-Bench, a benchmark suite for evaluating the performance of autonomous AI agents on real-world terminal-based tasks, have released version 2.0 alongside Harbor, a new framework for testing, improving and optimizing AI agents...