Free APILLMAPI keyOfficial

Cloudflare Workers AI

Run 40+ open models at the edge with a free daily Neuron budget

Workers AI serves open models (Llama, Mistral, Qwen, embeddings, image, ASR) from Cloudflare's edge, callable from a Worker or directly over the REST API using your account ID and an API token. The free plan includes a daily allocation measured in Neurons, a unit of GPU compute.

View docs Get an API key

Free tier

Free allocation of 10,000 Neurons/day on the Workers Free plan (no card to start). Above that, $0.011 per 1,000 Neurons on Workers Paid. Resets daily at 00:00 UTC.

Rate limits

Throttled by the 10,000 Neurons/day free budget rather than a fixed RPM. Neurons consumed per request depend on model and tokens; a ~500-token Llama 3 response costs roughly 400-600 Neurons.

Auth

API key

Base URL

https://api.cloudflare.com/client/v4/accounts/{account_id}/ai/run

What you'd build with it

Add inference to an edge Worker with zero cold-start round-trip to a separate provider
Free embeddings + Vectorize for a small RAG app inside the Cloudflare stack
Whisper transcription and image models alongside text in one platform
Tiny-volume automations comfortably inside 10k Neurons/day

Key endpoints

POST/@cf/meta/llama-3.1-8b-instructRun text generation on Llama 3.1 8B
POST/@cf/baai/bge-base-en-v1.5Generate text embeddings
POST/@cf/openai/whisperSpeech-to-text transcription

Members

Members get the runnable recipe

Create a free account Sign in