Cloudflare Workers AI
Run open models (LLMs, embeddings, image, speech) on Cloudflare's GPUs from a Worker or the REST API. · Cloudflare · AI
Overview
Workers AI runs open-weight models on Cloudflare's network. The free daily Neuron allowance is a hard cap on the Workers Free plan, so it cannot create a bill.
Good fit
Use when you already deploy on Cloudflare and want small, capped AI inference with no card.
- side projects
- edge apps
- embeddings
Not a fit
Avoid if you need frontier closed models or steady high volume; the daily cap is small.
- high-volume inference
- frontier models
Quickstart · 3 steps
- Create a free Cloudflare account.
- Create a Worker and add an AI binding.
- Call env.AI.run() with a model name.
Other free ai options
Free tierWhat you get freeRiskCard
GitHub · AI
2,000 code completions per month · 50 chat requests per month (including Copilot Edits) · Agent mode, CLI, and MCP servers included with limited usage
Billing risk: None
No card
Google Cloud · AI
Sessions run up to 12 hours, depending on availability · GPU/TPU access heavily restricted on the free tier · Exact resource limits are not published and vary over time
Billing risk: None
No card
Google Cloud · AI
Free input and output tokens on supported models · Limited access to certain models · 5,000 free Google Search grounding requests/month, shared across Gemini 3.x models
Billing risk: None
No card
Hugging Face · AI
Unlimited public repos (best-effort free storage) · 100GB private storage · Storage limits apply across models, datasets, and buckets
Billing risk: None
No card