harbor-index/harbor-index
Assistants
Reasoning
Harbor Index: 80 agentic evaluation tasks spanning SWE, security, science, math, and optimization (revision 1.4).
Run this task
CLI:
inspect eval inspect_harbor/harbor_index --model openai/gpt-5Python:
from inspect_ai import eval
from inspect_harbor import harbor_index
eval(harbor_index(), model="openai/gpt-5")Dataset information
| Harbor registry | harbor-index/harbor-index |
| Inspect task | harbor_index |
| Latest digest | sha256:74af9ee31f50d8dac213e67ebaeeddc6bb0e4470a28968f84766916e09fe13dd |
| Samples | 80 |
See Task Parameters for the parameter set shared across all Harbor tasks.