SpreadsheetBench

SpreadsheetBench: Towards Challenging Real World Spreadsheet Manipulation | Paper | Github | Homepage | We introduce SpreadsheetBench, a challenging spreadsheet manipulation benchmark exclusively derived from real-world scenarios, designed to immerse current large language models (LLMs) in the actual workflow of spread

Papers using SpreadsheetBench (5)

SkillOpt-Lite: Better and Faster Agent Self-evolution via One Line of Vibe2026SkillCAT: Contrastive, Assessment-Augmented and Topology-AwareSkill Self-Evolution for LLM Agents2026SkillAxe: Sharpening LLM-Authored Agent Skills Through Evaluation-Guided Self-Refinement2026Skills on the Fly: Test-Time Adaptive Skill Synthesis for LLM Agents2026ClawTrace: Cost-Aware Tracing for LLM Agent Skill Distillation2026