Skip to main content

Introduction

The openbench CLI provides a simple, powerful interface for running benchmark evaluations across 15+ model providers. All commands follow a consistent pattern and provide detailed feedback.

Command Structure

All openbench commands follow this pattern:
bench <command> [arguments] [options]

Available Commands

eval

Run a benchmark evaluation

eval-retry

Retry a failed evaluation

list

List all available benchmarks with descriptions

describe

Get detailed information about a specific benchmark

view

View previous evaluation results

Getting Help

# General help
bench --help

# Command-specific help
bench eval --help
bench describe --help