Product

Harbor

Open-source evaluation framework used to run, grade and rerun Terminal-Bench and Terminal-Bench-Science trials.

github.comCovered in 1 article · September 2026

Mentioned in

Sep 3, 2026

AIScientists Built a 70-Task Benchmark for AI Agents. Its Launch Leader Solved 30%.

Covered alongside

Ai2 Allen Institute Anthropic Argonne National Laboratory Bespoke Labs Claude Code