Reduce unexpected AI infrastructure costs by 30% and save 2 hours/week on manual cost tracking.
~2 hrs
Saved / wk
Medium
Setup
AI-Powered
AI
Every time your AI coding agent starts a new task, Freestyle launches a dedicated VM for it. Helicone (a tool that tracks AI API usage and costs) logs how many tokens (units of AI processing) each agent uses. You see exactly what each task costs before your bill surprises you.
~2 hrs
Saved / Week
Some setup required
Setup Complexity
AI Status
Step 1: Set up Freestyle to tag each VM with a task ID when it launches
Step 2: Route all AI model calls from your VMs through Helicone's proxy (a middleman that records traffic)
Step 3: Add the task ID as a custom property in Helicone so costs group by task
Step 4: Set up Helicone cost alerts to notify you when a single task exceeds your budget threshold
Step 5: Review the Helicone dashboard weekly to spot which agent tasks cost the most
Upgrade to Plus for AI-powered step-by-step guides
See PlansFreestyle gives you fast, full Linux virtual machines (VMs — isolated computers in the cloud) so your AI coding agents can build, test, and run code at scale. You can spin up a VM in under 600ms, clone it instantly, and pause it when idle so you only pay for what you use. It's built for developers who are shipping AI-powered tools and need reliable, scalable infrastructure behind them.