Ramp unveils free U.S. AI model routing service called Router
Expense-management firm Ramp introduced Router, an API that lets U.S. customers switch among several large-language models, free through 2026.
Ramp announced the rollout of Router, an API-based routing platform that enables businesses to select and alternate among a range of large-language models, including those from OpenAI, Anthropic, DeepSeek, Moonshot, Minimax, Nvidia, xAI and Z.ai. Available only in the United States, the service is offered at no charge through the end of 2026 and comes with a $26 launch credit, though users still pay for the underlying model inference fees.
Router provides configurable strategies, allowing users to prioritize cost tiers, benchmark performance, or reserve expensive models for complex tasks, and includes a monitoring dashboard that displays token spend, latency and fallback attempts. The platform adopts an opt-out retention policy, storing inputs, outputs and tool calls for one year after removing personally identifiable information. By entering the model-routing market, Ramp aims to capture a share of the growing AI inference sector and to integrate the service with its existing expense-management and token-monitoring products, potentially opening new client channels and partnerships with AI labs.
Why it matters
The launch gives businesses a cost-effective way to access multiple AI models and expands Ramp's foothold beyond expense management.
In this story
