Ramp, the fintech company behind corporate expense management, introduced Router on Wednesday—a new API-based platform enabling customers to seamlessly integrate and switch between large language models from providers including OpenAI, Anthropic, and DeepSeek. The company has employed Router internally for its own AI needs over the past three years and is now opening it to US-based users at no charge through the end of 2026, alongside a $26 launch credit, though 2027 pricing remains unannounced.

Router includes flexible routing strategies to help organisations optimise model selection—users can prioritise lower-cost inference tiers, rely on algorithm-chosen providers based on custom performance metrics, or route computationally demanding tasks exclusively to premium models. The platform provides visibility into spend, latency, and token usage through a dedicated dashboard, alongside a one-year default data retention policy with promised removal of personally identifiable information before using captured data to enhance the product. For Ramp, the move positions itself to capitalise on the expanding AI inference market whilst offering its existing customer base integrated model-switching capabilities within its broader expense-tracking ecosystem.