klinikebench/klinikebench

KlinikeBench: 333 long-horizon clinical consultations with an LLM-simulated patient, a 140-tool clinical action surface and four-gate scoring (diagnosis, required tools, parameterised actions, LLM-judged must-ask coverage).

← Back to Registry

Run this task

CLI:

inspect eval inspect_harbor/klinikebench --model openai/gpt-5

Python:

from inspect_ai import eval
from inspect_harbor import klinikebench

eval(klinikebench(), model="openai/gpt-5")

Dataset information

Harbor registry	klinikebench/klinikebench
Inspect task	klinikebench
Latest digest	sha256:202c363c2a6ecd1863fa43a52a0753a38e0fa7303c81e9b1ff891f9bb2087b23
Samples	333
Source	https://github.com/Zehui127/klinikebench

See Task Parameters for the parameter set shared across all Harbor tasks.