orinlabs/horizon-public
Assistants
Horizon: public example subset of a learning benchmark for extremely long-horizon agents — the agent must acquire knowledge from a long first-person session history and apply it to resolve a later request in the environment.
Run this task
CLI:
inspect eval inspect_harbor/orinlabs_horizon_public --model openai/gpt-5Python:
from inspect_ai import eval
from inspect_harbor import orinlabs_horizon_public
eval(orinlabs_horizon_public(), model="openai/gpt-5")Dataset information
| Harbor registry | orinlabs/horizon-public |
| Inspect task | orinlabs_horizon_public |
| Latest digest | sha256:38204787d8cd258d35e620e639c4418e4805141f22dc752b5e09431d0a3dd552 |
| Samples | 3 |
| Source | https://github.com/orinlabs/horizon |
See Task Parameters for the parameter set shared across all Harbor tasks.