Ramp launches AI model router to compete in inference market

What happened

Ramp, a corporate expense management platform, launched Router on Wednesday evening—a service that lets users and companies switch between large language models (AI systems that understand and generate text) from providers including OpenAI, Anthropic, DeepSeek, Moonshot, Minimax, Nvidia, xAI, and Z.ai through a single API. The service is free through the end of 2026 (though users pay for the underlying AI inference costs) and comes with a $26 credit launch offer; it is currently available only in the United States.

Why it matters

Ramp, which raised $750 million at a $44 billion valuation in June, is entering the growing AI inference market—the business of running AI models—while offering its existing customers a model routing tool that integrates with its current expense management products. By building relationships with AI labs and inference providers, Ramp may be able to acquire new customers and create additional entry points for selling its core services.

What to watch

Router offers several routing strategies, including one that lets users set preferences for model providers' flex usage tiers and another that chooses models based on up to three user-specified benchmarks. The company has not disclosed pricing for 2027 onward. It also retains model inputs, outputs, and tool calls for one year by default (removing personally identifiable information before using the data to improve the product), which users can opt out of.