Ramp Launches Router, an AI Model Routing Service for Enterprises

Ramp enters AI inference routing

Corporate expense management platform Ramp has launched Router, an AI model routing service that allows businesses to access and switch between multiple large language models through a single API. The company announced the service on Wednesday evening, positioning itself as a competitor to similar offerings from Stripe and OpenRouter.

Router provides access to models from OpenAI, Anthropic, DeepSeek, Moonshot, Minimax, Nvidia, xAI, and Z.ai. Ramp has been using the routing technology internally for the past three years to manage its own AI infrastructure needs before commercializing it.

The service is currently available only in the United States and will be free to use through the end of 2026, though users must still pay underlying model inference costs. Ramp is offering a $26 credit as a launch promotion but has not disclosed pricing for 2027.

Why it matters

Ramp's move into AI model routing represents a strategic expansion beyond its core expense management business. By offering Router alongside its existing AI token monitoring and spend management tools, the company creates a more comprehensive platform for enterprises managing AI costs. This positions Ramp to capture revenue from the rapidly growing AI inference market while deepening relationships with existing customers. If Router attracts significant usage, it could also establish Ramp as a key distribution channel for AI labs seeking enterprise customers.

Routing strategies and monitoring tools

Router offers several automated routing strategies designed to optimize cost and performance. Users can configure the service to route queries based on model providers' flexible usage tiers, select models according to up to three user-specified benchmarks, or direct only complex queries to expensive models while handling simpler requests with cheaper alternatives. The platform also enables easy model testing without manual switching.

A dashboard provides visibility into token spend, cost, latency, fallback attempts, and other operational metrics—capabilities that align with Ramp's existing expense management expertise.

Data retention raises questions

Router employs an opt-out data retention policy that may concern some enterprise users. By default, the service records model inputs, outputs, and tool calls for one year. Ramp states it will remove personally identifiable information before using this content to improve the product, but the opt-out default differs from the opt-in approach some enterprises prefer for sensitive data.

Strategic positioning

Ramp raised $750 million at a $44 billion valuation in June. The Router launch creates a two-pronged growth opportunity: tapping the AI inference market directly while offering existing clients an integrated service that complements Ramp's financial management tools. If Router becomes a popular testing ground for AI models—similar to OpenRouter's role in the developer community—Ramp could establish valuable relationships with AI labs and inference providers globally, creating new customer acquisition channels for its core products.

The details were first reported by TechCrunch.