Ramp, the corporate spend platform, has entered the model routing business with a service called Router. The API sends each AI request to the lowest-cost model that meets a developer’s quality bar, a technique the company says it has used internally for three years.
Router launched on August 19 and is free for the rest of 2026, with users paying only for the tokens they consume and a $26 credit for new signups. It routes to models from OpenAI, Anthropic, DeepSeek, Moonshot, MiniMax, Nvidia, xAI, and Z.ai, and offers strategies such as flex-tier preference, benchmark-based selection, and sending hard problems to premium models.
A dashboard tracks token spend, cost, latency, and fallback attempts. The service keeps model inputs, outputs, and tool calls for a year by default, though Ramp says it strips personally identifiable information and offers an opt-out.
Router enters a market heating up after Stripe’s $7B OpenRouter deal, giving developers another toll booth for AI inference traffic.
