Xpertbench: Expert Level Tasks with Rubrics-Based Evaluation

TL;DR

XpertBench is a comprehensive, expert-level benchmark with 1,346 tasks across multiple domains, designed to evaluate LLMs' proficiency in complex, real-world professional tasks using detailed rubrics and a novel LLM-based evaluation method.

Contribution

The paper introduces XpertBench, a high-fidelity benchmark with expert-derived tasks and ShotJudge, an innovative LLM evaluation paradigm to assess professional-level AI performance.

Findings

State-of-the-art LLMs achieve only ~66% success on XpertBench.

Models show domain-specific strengths and weaknesses.

XpertBench reveals a significant 'expert-gap' in current AI systems.

Abstract

As Large Language Models (LLMs) exhibit plateauing performance on conventional benchmarks, a pivotal challenge persists: evaluating their proficiency in complex, open-ended tasks characterizing genuine expert-level cognition. Existing frameworks suffer from narrow domain coverage, reliance on generalist tasks, or self-evaluation biases. To bridge this gap, we present XpertBench, a high-fidelity benchmark engineered to assess LLMs across authentic professional domains. XpertBench consists of 1,346 meticulously curated tasks across 80 categories, spanning finance, healthcare, legal services, education, and dual-track research (STEM and Humanities). These tasks are derived from over 1,000 submissions by domain experts--including researchers from elite institutions and practitioners with extensive clinical or industrial experience--ensuring superior ecological validity. Each task uses detailed rubrics with mostly 15-40 weighted checkpoints to assess professional rigor. To facilitate scalable yet human-aligned assessment, we introduce ShotJudge, a novel evaluation paradigm that employs LLM judges calibrated with expert few-shot exemplars to mitigate self-rewarding biases. Our empirical evaluation of state-of-the-art LLMs reveals a pronounced performance ceiling: even leading models achieve a peak success rate of only ~66%, with a mean score around 55%. Models also exhibit domain-specific divergence, showing non-overlapping strengths in quantitative reasoning versus linguistic synthesis.. These findings underscore a significant "expert-gap" in current AI systems and establish XpertBench as a critical instrument for navigating the transition from general-purpose assistants to specialized professional collaborators.