You.com introduces inference router with 12 open-weight models and 99.9% uptime guarantee

You.com has launched the Inference Router, a managed API service that enables developers to deploy open-weight models in production. The platform supports 12 live models, all hosted in the U.S., and offers a 99.9% uptime SLA along with zero data retention. It also includes server-side web search and pay-as-you-go pricing without platform fees. The Inference Router addresses challenges developers face during model migrations, such as security reviews and compliance with data privacy regulations. It routes requests first to You.com’s own inference capacity, failing over to vetted partners like Parasail, Wafer, Alibaba Cloud, Meta, and Inception if needed. All models are US-hosted and meet zero data retention standards. You.com claims its service offers cost advantages, with models like DeepSeek V4.1 Flash being 77% cheaper per question than Claude Sonnet 5.5. The platform also highlights improvements in benchmark tests, such as a 30% reduction in tool calls when using Nemotron 3.5 Lightning with You.com search.