Ramp Launches AI Model Router Service Free Through End of 2026
Corporate expense management platform Ramp launched Router on Wednesday, an AI model routing service available free for the rest of 2026 with a $26 credit offer at launch.

👁 122 reads
Ramp, a corporate expense management platform, has expanded into AI infrastructure with the launch of Router, a new AI model routing service announced on Wednesday evening. This service allows users and companies to access and switch between multiple large language models through a single API, eliminating the need to manage separate integrations for each AI provider. Router is being offered free of charge through the end of 2026, though users will still need to pay the underlying inference costs charged by the AI model providers themselves. To help users get started, Ramp is providing a $26 credit at launch. The company has not yet announced what pricing will look like when 2027 arrives, and the service is currently available only to users in the United States.
- Router gives users API access to models from OpenAI, Anthropic, DeepSeek, xAI, and others, with benchmark-based and tier-based routing strategies
- The service is free to use through the end of 2026 - users still pay underlying inference costs - and comes with a $26 launch credit
- Ramp, which raised $750 million at a $44 billion valuation in June, has used the technology internally for three years
What AI Models and Features Does Router Support
Router provides access to a comprehensive range of AI models from major providers including OpenAI, Anthropic, DeepSeek, Moonshot, Minimax, Nvidia, xAI, and Z.ai. This multi-provider approach means developers and businesses can work with different models without building separate integrations for each one. The service offers several intelligent routing strategies designed to optimize performance and cost. One strategy automatically selects the most appropriate model based on up to three user-specified benchmarks, allowing companies to prioritize factors like accuracy, speed, or cost-effectiveness. Another routing option targets providers’ flex usage tiers, which can help reduce costs during periods of variable demand. A third strategy reserves more expensive, powerful models only for queries that are identified as difficult or complex, while routing simpler requests to more economical options.
The platform includes a comprehensive dashboard that gives users visibility into their AI usage patterns. This dashboard surfaces important metrics including token spend, cost per request, latency measurements, and fallback attempts when a primary model is unavailable. This level of transparency helps organizations understand and control their AI spending while monitoring performance across different models and use cases.
Data Retention and Privacy Considerations
By default, Router retains model inputs, outputs, and tool calls for a period of one year. Ramp states that it removes personally identifiable information from this data before using it to improve the product. This data retention policy is important for organizations to understand when evaluating whether Router fits their compliance and privacy requirements. Companies handling sensitive information should review these terms carefully and consider whether the default retention settings align with their data governance policies.
Ramp’s Background in AI Infrastructure
According to Ramp, the company developed the underlying routing technology for its own internal AI needs over the past three years before deciding to offer it as a standalone product. This positions Router as a tool that emerged from real-world use rather than being built speculatively. Ramp raised $750 million at a $44 billion valuation in June, demonstrating significant investor confidence in the company’s direction. The company already offers AI token usage monitoring and spend management tools as part of its core business, and Router fits naturally alongside these existing products. This background in expense management and cost monitoring gives Ramp a unique perspective on the financial aspects of AI deployment, which is reflected in Router’s focus on cost optimization and spending visibility.
Google · Your sourcesQuantslant higher in your Google - one clickAdd to preferred sources →Frequently Asked Questions
Is Router completely free to use through 2026?
Router itself is free to use through the end of 2026, meaning Ramp does not charge a service fee for accessing the routing platform. However, users still need to pay the underlying model inference costs charged by the AI providers themselves, such as OpenAI or Anthropic. Ramp provides a $26 credit at launch to help offset these costs initially.
Which AI model providers can I access through Router?
Router provides access to models from OpenAI, Anthropic, DeepSeek, Moonshot, Minimax, Nvidia, xAI, and Z.ai. All of these providers can be accessed through a single API, eliminating the need to manage separate integrations for each one.
Where is Router available?
Router is currently available only in the United States. Ramp has not announced plans for international availability at this time.
More on this from QUANTSLANT
magazine·Greg Brockman Takes Day-to-Day Control at OpenAI as Executives DepartOpenAI president Greg Brockman has quietly expanded his authority over product, commercial operations and scaling as a wave of senior executives departed throughout 2026.
