About Cloudflare Workers AI
Workers AI runs machine learning models on serverless GPUs across the Cloudflare network. You call the models from Workers, Pages or any code that can reach the Cloudflare API.
What it does
The model catalog includes more than fifty open-source models. You pay for what you use rather than for idle servers. The service sits alongside other Cloudflare products such as AI Gateway and Vectorize, so you can build retrieval and routing into one stack.
Key features
- Over 50 open-source models in the catalog
- Calls from Workers, Pages or the REST API
- Serverless GPUs with pay-for-use billing
- Beta options such as batch processing and function calling
- Works with AI Gateway and Vectorize
Who it's for
- Developers already building on Cloudflare Workers
- Teams that want AI features running close to their users
- Builders who want a single developer platform for AI and storage
Pricing and good to know
The service is available on both Free and Paid plans, with usage-based pricing for the models. Limits and rates are listed in the platform documentation. Some features are still in beta, so check their status before relying on them for production traffic.
Highlights
- Run over 50 open-source models
- Call models from Workers or Pages
- Pay for the compute you use
- Try beta batch and function calling
- Connect with AI Gateway and Vectorize
