What is Cloudflare Workers AI used for?

Cloudflare Workers AI is a serverless service for running machine learning models on Cloudflare's network using serverless GPUs. It is a good fit for adding inference to Cloudflare Workers apps, prototyping with hosted open models and batch processing prompts asynchronously.

How much does Cloudflare Workers AI cost?

Cloudflare Workers AI is a paid product with no free plan. Usage-based pricing per million tokens, varying by model. Llama 3.2 1B costs $0.027 per million input tokens and $0.201 output; embeddings start at $0.012 per million input tokens.

Is Cloudflare Workers AI open source?

No. Cloudflare Workers AI is proprietary (closed-source) software and can't be self-hosted. Open-source alternatives to Cloudflare Workers AI include beta9, vLLM and LocalAI.

What are some alternatives to Cloudflare Workers AI?

Cloudflare Workers AI competes with Replicate, Together AI and Fireworks AI. For open-source options, see Enlisted's ranked list of open-source Cloudflare Workers AI alternatives.

Cloudflare Workers AI: features, pricing and alternatives

About Cloudflare Workers AI

Workers AI lets developers run machine learning models on Cloudflare's network without managing servers or GPUs. Models are called from Cloudflare Workers code through bindings or over a REST API, and the service is billed as part of the wider Cloudflare platform.

Its documentation covers a model catalog, OpenAI-compatible API endpoints, integration with the Vercel AI SDK, an asynchronous batch API, function calling, JSON mode, prompt caching and fine-tuning with LoRA adapters. A markdown conversion feature turns supported file formats into text for model input. Related pages describe limits, pricing and a playground for trying models.

Workers AI is a proprietary, cloud-only service from Cloudflare, so there is no self-hosted option. Pricing and platform limits are documented in the Workers AI docs rather than summarized on the overview page.

Key features

Serverless GPU inference without server management
Workers bindings and REST API access
OpenAI-compatible API endpoints
Function calling and JSON mode
LoRA adapter fine-tuning support
Asynchronous batch API

Good fit for

→Adding inference to Cloudflare Workers apps
→Prototyping with hosted open models
→Batch processing prompts asynchronously

Tags: serverless; inference; gpu; cloudflare; llm; edge; ai-infrastructure

Cloudflare Workers AI: questions and answers

What is Cloudflare Workers AI used for?: Cloudflare Workers AI is a serverless service for running machine learning models on Cloudflare's network using serverless GPUs. It is a good fit for adding inference to Cloudflare Workers apps, prototyping with hosted open models and batch processing prompts asynchronously.
How much does Cloudflare Workers AI cost?: Cloudflare Workers AI is a paid product with no free plan. Usage-based pricing per million tokens, varying by model. Llama 3.2 1B costs $0.027 per million input tokens and $0.201 output; embeddings start at $0.012 per million input tokens.
Is Cloudflare Workers AI open source?: No. Cloudflare Workers AI is proprietary (closed-source) software and can't be self-hosted. Open-source alternatives to Cloudflare Workers AI include beta9, vLLM and LocalAI.
What are some alternatives to Cloudflare Workers AI?: Cloudflare Workers AI competes with Replicate, Together AI and Fireworks AI. For open-source options, see Enlisted's ranked list of open-source Cloudflare Workers AI alternatives.