Computational Arbitrage in AI Model Markets

TL;DR

This paper explores how arbitrage strategies can be used in AI model markets to profit from model cost differences, impacting market competition, pricing, and entry of new providers.

Contribution

It introduces the concept of arbitrage in AI model markets, demonstrating its economic viability and analyzing its effects on market dynamics and provider revenues.

Findings

Arbitrage strategies can achieve up to 40% profit margins.

Arbitrage reduces market segmentation and lowers consumer prices.

Distillation can create significant arbitrage opportunities, affecting model revenue.

Abstract

Consider a market of competing model providers selling query access to models with varying costs and capabilities. Customers submit problem instances and are willing to pay up to a budget for a verifiable solution. An arbitrageur efficiently allocates inference budget across providers to undercut the market, thus creating a competitive offering with no model-development risk. In this work, we initiate the study of arbitrage in AI model markets, empirically demonstrating the viability of arbitrage and illustrating its economic consequences. We conduct an in-depth case study of SWE-bench GitHub issue resolution using two representative models, GPT-5 mini and DeepSeek v3.2. In this verifiable domain, simple arbitrage strategies generate net profit margins of up to 40%. Robust arbitrage strategies that generalize across different domains remain profitable. Distillation further creates strong arbitrage opportunities, potentially at the expense of the teacher model's revenue. Multiple competing arbitrageurs drive down consumer prices, reducing the marginal revenue of model providers. At the same time, arbitrage reduces market segmentation and facilitates market entry for smaller model providers by enabling earlier revenue capture. Our results suggest that arbitrage can be a powerful force in AI model markets with implications for model development, distillation, and deployment.