Ramp launches Router, a free multi-brand AI service until 2026

Ramp introduces Router, a service that allows users to switch between multiple language models via an API and a tracking dashboard. The offering is currently limited to the United States and is free until the end of 2026, with a launch credit of $26. By default, Router records inputs, outputs, and tool calls for one year, with an opt-out option and a promise to delete identifying data. The service aggregates models from several providers and offers routing strategies based on user preferences.

Router retains exchanges for one year, with an opt-out option

By default, Router records inputs, outputs, and tool calls related to the models for one year, with the possibility to opt out. Ramp states that it will remove personally identifiable information before using this content to improve the product.

Free until the end of 2026 in the United States, $26 credit

The service is currently only available in the United States. It is free until the end of 2026, but users remain responsible for the inference costs of the AI models. A launch offer provides $26 in credit. Ramp has not specified the pricing that would apply next year.

Supported providers and model selection strategies

Router provides access to models offered by OpenAI, Anthropic, DeepSeek, Moonshot, Minimax, Nvidia, xAI, and Z.ai. The service offers several strategies to route requests based on customer preferences, including one that favors flexible usage levels among providers and another that automatically selects the model based on up to three criteria specified by the user. Clients can reserve challenging requests for expensive models or test models without changing configurations, and they have a dashboard to track token spending, costs, latency, and fallback attempts.

Positioning against OpenRouter, internal use, and prospects

Router operates similarly to OpenRouter, although the latter offers more model options than Ramp's current offerings. The company claims to have been using this router internally for three years. This new component complements its existing solutions for tracking token usage and controlling associated spending. In June, Ramp raised $750 million, bringing its valuation to $44 billion. If Router proves to be as relevant as OpenRouter for model evaluation, Ramp could establish long-term partnerships with AI labs and inference providers, potentially facilitating the acquisition of new clients and providing new access to its expense management tools.