Ramp Launches Router to Cut Enterprise AI Costs

Companies deploying generative AI face a growing operational burden as model prices, performance and availability shift across providers. Integrating each vendor separately can lock developers into fragmented infrastructure and leave finance teams with limited visibility into token spending. Ramp, best known for its corporate expense-management platform, is turning a routing system used internally for three years into a commercial service that selects models according to quality, cost and latency through a single API.

Ramp launched Router.com in the United States on Aug. 19, 2026, offering access to models from OpenAI, Anthropic and other providers, along with custom routing, automatic failover and dashboards for tracking token usage, cost and latency. Ramp said the system routes more than 2.75 trillion tokens a month and has lowered its own inference costs by about 30%, while early customers have saved 40% on average. Routing is free through the end of 2026, and new users receive $26 in promotional credits while paying list prices for model tokens.