These tools and metrics are designed to help AI actors develop and use trustworthy AI systems and applications that respect human rights and are fair, transparent, explainable, robust, secure and safe.
ModelBench
ModelBench is an open-source tool that supports the safety and robustness of AI systems by providing a standardised methodology for benchmarking the safety performance of AI language models throughout the AI system lifecycle, particularly at the testing and evaluation phase.
Developed by MLCommons' AI Risk & Reliability Working Group, the tool operationalises safety assessment by running a defined battery of test prompts against an AI system under test, using automated annotator models to classify responses across a taxonomy of hazard categories (including violent content, hate speech, and facilitation of self-harm). Individual hazard scores are aggregated into system-level benchmark ratings and published as structured, human-readable reports, enabling comparison of relative safety performance against reference baselines.
Auto-discovered on 2026-09-09 by OECD Catalogue Automation
About the tool
You can click on the links to see the associated tools
Tool type(s):
Objective(s):
Purpose(s):
Lifecycle stage(s):
Type of approach:
Maturity:
Usage rights:
License:
Target users:
Risk management stage(s):
Technology platforms:
Programming languages:
Github stars:
- 133
Github forks:
- 35
Use Cases
Would you like to submit a use case for this tool?
If you have used this tool, we would love to know more about your experience.
Add use case
Partnership on AI
