Truth that Matters. Stories that Impact

Truth that Matters. Stories that Impact

Technology

Ramp Launches AI Model Routing Service Called Router for US Businesses

Corporate expense management platform Ramp has entered the artificial intelligence infrastructure space with the rollout of Router, an API-driven model routing service designed to let businesses switch between different large language models.

What Happened

Ramp announced the launch of Router on a Wednesday evening, introducing a tool that allows companies to access and alternate among multiple large language models via a single API. According to Ramp, the firm has utilized this proprietary router internally to handle its own AI workflows over the past three years. The service is currently limited to users based in the United States. Ramp is offering the routing layer for free through the remainder of 2026 alongside a $26 credit launch promotion, though customers remain responsible for underlying model inference expenses. The company has not yet detailed its pricing structure for the following year.

Key Highlights

  • Supported Providers: Router connects users to models developed by OpenAI, Anthropic, DeepSeek, Moonshot, Minimax, Nvidia, xAI, and Z.ai, functioning along lines similar to OpenRouter, which maintains a broader catalog.
  • Custom Routing Strategies: Organizations can direct queries using specific parameters, such as prioritizing flex usage tiers, selecting models using up to three custom benchmarks, or reserving expensive models exclusively for complex tasks.
  • Monitoring Dashboard: The platform includes analytics for tracking token expenditures, overall costs, response latency, and fallback attempts.
  • Data Retention Policy: The tool uses an opt-out logging policy, retaining model inputs, outputs, and tool calls for one year by default, while removing personally identifiable information prior to using data for system improvements.
  • Corporate Valuation: Ramp previously secured $750 million in funding at a $44 billion valuation in June.

Why This Matters

The introduction of Router provides Ramp with an opportunity to capitalize on the growing AI inference sector alongside competitors like Stripe. The service aligns directly with Ramp’s existing financial management software, which already includes features for tracking AI token consumption and spending. Providing a routing and testing environment also positions Ramp to establish partnerships with global AI labs and inference providers, creating a pathway to attract new corporate clients to its core expense management offerings.

What to Watch Next

Ramp has not yet announced its post-2026 pricing model or timeline for expanding Router beyond the United States. Observers can watch for future updates regarding fee structures, international availability, and whether Ramp integrates additional AI model providers into the platform.

Frequently Asked Questions

What is Ramp Router?

Router is an API-based service launched by Ramp that allows enterprises and developers to switch between various artificial intelligence models based on cost, latency, or specific technical benchmarks.

Which AI models are supported on the platform?

The service currently offers models from OpenAI, Anthropic, DeepSeek, Moonshot, Minimax, Nvidia, xAI, and Z.ai.

Is Router free to access?

The routing service is free through the end of 2026 and includes a $26 launch credit, but customers must still cover the standard inference fees charged by model providers.

How does Ramp handle user data on Router?

Router retains tool calls, inputs, and outputs for one year under a default opt-out policy, with Ramp stating it removes personally identifiable information before utilizing data for product development.

Source: TechCrunch