Ramp Launches Router, a Model-Switching Service for AI Inference

Ramp, a corporate expense management platform, is extending its foothold in AI infrastructure by unveiling Router, a new service that functions as a routing layer for large language models. The platform allows users to access and switch between multiple AI models via a unified API interface, following a similar approach to rival services like OpenRouter.

Service Details and Availability#

Router launches initially in the United States at no cost through the end of 2026, with users paying only for actual model inference consumption. New users receive a $26 launch credit, though Ramp has not disclosed its pricing structure beyond this year.

The routing service provides access to models from OpenAI, Anthropic, DeepSeek, Moonshot, Minimax, Nvidia, xAI, and Z.ai.

Smart Routing Capabilities#

Ramp built Router to handle increasingly complex AI usage patterns. The service includes multiple routing strategies that automatically direct requests based on customizable preferences:

• Route based on provider flex usage tier pricing
• Automatically select models using up to three user-defined performance benchmarks
• Direct only computationally expensive queries to premium models
• Test new models without requiring manual switching

Users access a control dashboard showing token expenditure, inference latency, costs, fallback activity, and related metrics.

Data Retention and Privacy#

By default, Router retains model inputs, outputs, and tool calls for one year for product improvement purposes. Ramp says it removes personally identifiable information before using this data internally, and offers an opt-out mechanism for those seeking stricter privacy controls.

Strategic Implications for Ramp#

Ramp has been operating an internal router for its own AI needs over the past three years, positioning the company to offer Router as a complementary service alongside its existing AI token monitoring and spend management tools. This move opens multiple revenue streams: direct routing fees, increased engagement with current clients, and potential relationships with AI model providers and inference companies worldwide.