Cloudflare Workers AI Runs Large Models — Kimi K2.5 First Up

Cloudflare Blog·medium signal

Cloudflare announced Workers AI now supports large-scale model inference, with Moonshot AI's Kimi K2.5 as the first large model available on the platform. The company optimized its inference stack specifically to reduce costs for internal agentic use cases, enabling developers to power long-running agent workflows entirely within Cloudflare's edge network. This shifts Cloudflare from a CDN/security provider into a full-stack AI inference layer.