cloudflare-workers-ai

Cloudflare Workers AI enables you to run AI models at the edge by binding the AI service in wrangler.jsonc and calling env.AI.run(model, inputs). This lets you execute text generation, embeddings, image generation, and vision models directly on Cloudflare's infrastructure without routing requests to external APIs.

Yes, Cloudflare Workers AI supports streaming for text generation via the AI binding. Enable streaming in your inputs to receive LLM responses incrementally, reducing latency and improving user experience for long-form text output.

Use Cloudflare Workers AI to generate embeddings for your documents, then retrieve and rank them for retrieval augmented generation. The AI binding handles embedding model inference at the edge, enabling fast semantic search without external vector databases.

Yes, Cloudflare Workers AI provides OpenAI-compatible endpoints, allowing you to use the same client libraries and API patterns you're familiar with while running models at the edge with AI Gateway integration for logging and cost tracking.

Cloudflare Workers AI supports text generation, embeddings, image generation, vision models, and function calling. It integrates with AI Gateway for request logging, cost tracking, and rate limit resilience while handling model availability automatically.

Integrate Cloudflare AI Gateway with Workers AI to centralize logging, cost tracking, and analytics. Gateway captures all inference requests, providing visibility into model usage and spending across your edge applications.