E-Commerce Bench: A New Benchmark for Long-Horizon Autonomous Business Operations

Qwen ·

Key Info

E-Commerce Bench is a new benchmark for long-horizon autonomous business operations, in which AI agents start with ¥100,000 to run online stores for 365 days, managing sourcing, negotiation, pricing, promotions, inventory, and cash flow in a realistic e-commerce market.

Highlights

  • Grounded in real economics: 6,886 products, 576 suppliers (including 152 fraudsters), 600-minute workdays, storage fees, returns, and reputation systems.
  • Designed to test long-horizon learning: almost no model learns to buy cheaper or improve its strategy across the simulated year.
  • Covers diverse business tasks, from supplier negotiation to cash-flow management, making it a strict stress test for agentic AI.
Loading...