Reflection AI's Beam gets limited-time free access on Vercel AI Gateway
Vercel describes Beam as a low-latency US open-weight model that previously ran in stealth as stealth/cat-jam.
TLDR
On October 16, 2026, Vercel announced that Reflection AI's Beam was free on its AI Gateway for a limited time. It described Beam as a low-latency US open-weight model that had run in stealth as stealth/cat-jam. Anyscale says users can run Beam at scale in their own accounts with Ray Serve LLM on Anyscale, without managing infrastructure.
Combined views
10.2K
5 Sources, first seen ago
