Corporate expense management platform Ramp is moving quickly to establish its own toll booths for AI inference, following closely in the footsteps of Stripe. On Wednesday evening, the company unveiled its own AI model routing service called Router, which enables users and businesses to access and toggle between various large language models through a single API. According to Ramp, it has been using this internally built router to manage its own AI needs for the past three years. The service is currently available only in the United States, and it will remain free to use through the rest of 2026—though users will still need to cover the underlying AI model inference costs. As part of the launch, Ramp is also offering a $26 credit. The company hasn't disclosed what the pricing will look like once the free period ends.

In terms of functionality, Router operates much like OpenRouter, although OpenRouter provides access to a far wider selection of AI models. Router currently supports models from OpenAI, Anthropic, DeepSeek, Moonshot, Minimax, Nvidia, xAI, and Z.ai. It also includes several routing "strategies" that allow customers to direct AI requests based on their own preferences. For instance, one strategy lets users prioritize model providers' flexible usage tiers, while another allows Router to pick the most suitable model based on up to three user-defined benchmarks. Users can also choose to send only the most difficult queries to premium models, or easily experiment with different models without needing to switch manually. A dashboard is included as well, offering visibility into token spend, costs, latency, fallback attempts, and other key metrics. One noteworthy detail: Router has an opt-out data retention policy. By default, it will log model inputs, outputs, and tool calls for a full year, though the company says it will strip out personally identifiable information before using that data to improve the product.

For Ramp, moving into model routing represents a two-sided opportunity. On one hand, it allows the company to tap into the rapidly expanding AI inference market. On the other, it gives existing clients a routing service that complements Ramp's current suite of tools, which already includes AI token usage monitoring and spend management. If Router becomes as popular a testing ground for models as OpenRouter has turned out to be, Ramp could also build durable relationships with AI labs and inference providers around the world. That, in turn, could help the company—which raised $750 million at a $44 billion valuation in June—attract new customers and create another entry point for selling its expense management products.