You.com launched Inference Router, an OpenAI-compatible API that serves open-weight models and web search to teams running agents in production. The initial lineup includes Kimi K3, GLM 5.3, DeepSeek V4.1 Flash, Qwen 3.8 Flash, Muse Spark 1.3, and Mercury 3. Calls run on US-hosted infrastructure with zero data retention across model and search requests, under an uptime SLA of at least 99.9% backed by service credits. Companies are moving classification, extraction, summarization, and routine agent steps onto open-weight models, which now cover a growing share of production work at a fraction of the cost of frontier models. An uptime SLA and US hosting have usually required a committed contract, and open-weight hosts rarely include web search. Inference Router offers those controls on demand. Traffic goes first to You.com's own inference capacity and fails over to vetted partners, including Parasail, Inception, Meta, and Alibaba Cloud, when needed to hold the SLA. Search runs on the same server as the model, which You.com says reduced latency by up to 15% in internal tests. There is no platform fee. Pricing is published per model on a pay-as-you-go basis, with volume discounts as usage scales. You.com was founded by Richard Socher, its CEO and a former chief scientist at Salesforce, and Bryan McCann. The company supplies web search APIs for AI agents and has run models together with live web results since December 2022. Inference Router adds the model layer so a team can ground an agent and run inference through one API, by changing the base URL in existing OpenAI client code.