Workers AI

Serverless AI inference: deploy 50+ open-source models (like Llama and Mistral) on Cloudflare's global GPU network for low-latency edge execution.

Workers AI delivers fast, affordable AI inference-as-a-service, running models directly on Cloudflare's global network of serverless GPUs. You invoke models from your code (Workers, Pages, or API) without managing infrastructure or paying for idle compute. This platform supports a curated catalog of 50+ popular open-source models—including Llama and Stable Diffusion—ensuring developers avoid vendor lock-in. Workers AI seamlessly integrates with Cloudflare's developer ecosystem: Vectorize (vector database), R2 (object storage), and AI Gateway (monitoring and control), providing a unified, efficient, full-stack solution for building real-time GenAI applications.

https://developers.cloudflare.com/workers-ai/