Ramp

Fintech · AI Infrastructure · Enterprise Software

Launch

Ramp launches Router.com to cut AI inference costs by 40%

AI spending at companies has ballooned nearly 21-fold in a year, and most finance teams still can't see where it goes.

• Ramp launched Router.com on August 19, 2026, a single API that sends each AI request to the lowest-cost model meeting a set quality bar, with customers already cutting inference costs 40% on average.
• Ramp is a corporate spend-management platform that processes more than $200 billion in purchases annually and is extending that cost-control approach to AI usage.
• Router connects developers to closed models from OpenAI, Anthropic and SpaceXAI (formerly xAI) alongside open-weight models like DeepSeek, Qwen and Kimi served via providers such as Fireworks AI, with Gemini support coming soon.
• Router already handles more than 2.75 trillion tokens monthly, and Ramp says the same system cut its own internal inference costs about 30% while keeping reliability above 99.9%.
• The launch follows Ramp's own AI Index showing customer AI spending has grown 20.7x since June 2025, making it one of the fastest-rising line items on corporate books.
• Routing is free through the end of 2026, with new users getting $26 in model credits and a limited-time 50% discount on GPT-5.6 Sol.
• By tying model routing directly to spend visibility, Ramp is betting that AI costs become a controllable budget line rather than an opaque, fast-growing expense, a shift that could reshape how enterprises buy AI compute.