Building a Self-Hosted AI Agent on Cloudflare's Free Tier: From Zero to Production

Description and use

Building a self‑hosted AI agent on Cloudflare’s free tier using Ollama and Workers to avoid costly cloud services.

Problem

Deploying AI agents on expensive cloud infrastructure is costly; need a free, production‑ready solution.

Outcome

The agent runs entirely on free Cloudflare infrastructure with a minimal VPS cost, achieving sub‑50 ms edge latency and <$0.01 inference cost per 1,000 tokens.

Workflow

• Spin up a $5–$6/month VPS with 4 GB RAM and install Ollama. Run curl https://ollama.ai/install.sh | sh and pull the quantized Mistral 7B model. Expose Ollama via Cloudflare Tunnel: download cloudflared, login, create a tunnel, and route DNS to ollama.yourname.workers.dev. Create a Cloudflare Workers project with npm create cloudflare@latest my-ai-agent -- --type javascript. Replace src/index.js with a JavaScript agent that classifies intent and routes to search, calculate, or chat handlers, all calling the Ollama endpoint. Deploy the Worker and test with a POST request to the Worker URL. Monitor usage within Cloudflare’s free tier limits.

Evidence

Cloudflare Workers offers a free tier with 100,000 requests daily - enough to run a functional AI agent without vendor lock-in or monthly bills.

Source

Ollama runs quantized LLMs locally. Start by spinning up a small VPS (Linode, Hetzner, or DigitalOcean at $5–$6/month) with 4GB RAM and install Ollama:

Source

A $5 VPS plus free Cloudflare Tunnel replaces $30–$50/month managed inference APIs.

Source