Dispatch
Ramp turns three years of internal AI plumbing into Router, a model-routing API free through 2026
Ramp launched Router on 20 August, exposing the model routing service it built internally over three years as a public API covering OpenAI, Anthropic, DeepSeek, Moonshot, Minimax, Nvidia, xAI and Z.ai. Routing strategies include preferring flex usage tiers, specifying up to three performance benchmarks for automatic selection, or sending only complex queries to expensive models. It is free through the end of 2026 with a $26 launch credit (you still pay inference costs), US-only, and 2027 pricing is unannounced.
Source
↳ Follow the thread