Competition in the retail AI space is heating up.

Alibaba.com released the results from a 107-task benchmark evaluation revealing that Accio, the company’s AI agent platform for e-commerce, completed a variety of tasks at more than 50 percent lower estimated cost as compared to OpenAI Codex and Anthropic’s Claude Code, and it did so with comparable completion quality.

“This makes Accio the most cost-competitive e-commerce AI tool available to small and medium-sized businesses,” Alibaba.com said in a statement Wednesday, adding that it also introduced an expanded Accio “that brings research, sourcing and daily e-commerce operations into one AI workspace for small businesses.” This saves time and money.

The company made the announcement at CoCreate, Alibaba.com’s annual event for entrepreneurs and small and medium-sized businesses (SMBs).

Kuo Zhang, president of Alibaba.com, said for small businesses, “unaffordable AI is useless. Our goal is not simply to make AI more powerful. It is to make commerce AI practical and affordable for even a one-person company.”

Regarding the benchmark report, Alibaba said Accio’s efficiency comes from optimizing the full system around real commerce work. This includes what the company describes as “commerce-specific intelligence.”

“Accio uses commerce-specific data and workflows to post-train lightweight models for clearly defined tasks,” the report’s authors said. “This allows smaller models to handle more routine work efficiently, while more capable models remain available when deeper reasoning is required.”

Instead of choosing the cheapest model or default to the most powerful one, the platform allocates the right resources for each task. “It breaks complex requests into manageable steps and balances quality, speed, cost and data requirements for each step,” the report noted. For SMBs, the report’s authors said all of this means getting the performance that the job requires without paying for any computing it doesn’t need.

Alibaba also said Accio has now evolved into a unified workspace for global e-commerce operations. The report’s authors said SMBs can use the platform to research markets and identify product opportunities as well as develop products. Accio can also evaluate suppliers and manage daily operations.

And through links with Amazon, Shopify, eBay, TikTok Shop and Walmart, Alibaba said sellers can access supported storefront workflows from a single place.

In addition, Alibaba said it has now open-sourced Commerce Agent Bench on GitHub. Instead of relying on synthetic exercises, the company said it draws on real merchant activity, including 10 million active SMB users, 1.6 million conversations, and 200,000 execution traces. All of which are distilled into 107 end-to-end commerce tasks across seven categories and four levels of autonomy.

“Tasks include reviewing hundreds of unstructured emails, spotting payment fraud, calculating landed costs and booking multi-carrier shipping routes,” the company said. “No single model led across the board, reinforcing the case for task-level routing—the idea that different tasks are best handled by different AI models, rather than relying on one general-purpose model for everything.”