Case Study: Poolside launches its Laguna models in record time with Baseten

Poolside launches Laguna in 7 weeks with Baseten, hitting 146 ms P50 TTFT

Poolside, a frontier AI research company, needed to launch its Laguna model family to developers with a branded, production-grade API. The challenge was to provide a white-labeled, high-performance inference platform with billing, usage controls, and secure API key federation without having to build the operational infrastructure themselves. They partnered with Baseten to meet a tight deadline.

Baseten implemented a complete inference platform for Poolside, including dedicated model hosting with an OpenAI-compatible API, KV cache-aware routing, and integrated billing webhooks. The solution was deployed in just seven weeks from kickoff to production. Baseten's work resulted in exceptional performance, with P50 time-to-first-token as low as 146 milliseconds, allowing Poolside to successfully launch its models on schedule.

Poolside