Corporate expense management provider Ramp has released Router.com, a multi-model routing service and API gateway designed to dynamically direct enterprise inference requests across commercial and open-weight large language models.
The launch introduces a unified endpoint connecting developer workflows to models from OpenAI, Anthropic, and SpaceXAI (Grok), with scheduled integrations for Google Gemini as well as open-weight families including DeepSeek, Kimi, Minimax, and Nvidia served through infrastructure providers such as Fireworks AI, Together AI, Baseten, and Crusoe.
Routing Architecture and Spend Governance
Rather than operating purely as a proxy layer, Router integrates model selection mechanics with Ramp's internal accounting and token-spend tracking systems. Enterprise teams can configure dynamic routing policies based on cost ceilings, latency constraints, provider fallback triggers, or evaluation benchmarks.
Ramp routes queries using an internal evaluation suite dubbed Ramp SWE-Bench, which continuously tests model versions against real production tasks to determine task-specific performance and cost efficiency. The platform also offers configurable strategies, including routing complex prompts to frontier models while shunting routine queries to smaller, low-cost options, and taking advantage of providers' off-peak or flex-pricing tiers.
According to Ramp CTO Rahul Sengottuvelu, the company developed the routing engine internally over three years to optimize its own production workloads, reporting an internal inference cost reduction of roughly 30% alongside 99.9% uptime across production pipelines.
Pricing and Data Retention Terms
Router is available to developers in the United States with free routing layer access through 2026 and a $26 introductory usage credit. Customers pay baseline list prices for the raw tokens consumed by underlying model providers.
Under its default terms of service, Router implements a one-year data retention policy that records user inputs, outputs, and tool calls for service improvements, with automated redaction of personally identifiable information. Organizations can opt out of data retention through their dashboard settings.
The launch reflects growing competition among fintech platforms to capture inference traffic and corporate AI budgets, following Stripe's recent acquisition of OpenRouter. By embedding model routing within its financial control suite, Ramp aims to convert API mediation into a direct extension of corporate spend management.
