PRICING
MODELS GITHUB
Pricing

Free.
Always.

Loophole is open-source. No subscriptions, no seats, no hidden fees. Pay only what you owe directly to your AI provider — we take nothing.

 
Community
$0
/ month
Everything you need to build, shipped free. Download, run, and never pay us a cent.
Download Free
Most popular
Pro
$0
/ month · same as community, for now
All the same features as Community, with priority access to new releases and Pro-tier features as we ship them.
Download Free
Coming soon
Team
TBD
/ seat / month
Shared infrastructure, centralised keys, and team-wide context sync on top of the free open-source core.
Join Waitlist
Feature breakdown
What's included
Feature Community Pro Team
Core Editor
Inline diffs & approval controls
Plan & Execution modes
VS Code extension support
AI & Models
Bring your own API key
Local model endpoints (Ollama, vLLM)
Centralised API key management
Team & Collaboration
Shared rules & skill libraries
Usage analytics per seat
Priority support
Latest
Loophole 2.0
Loophole 2.0: The Fastest AI IDE Yet
07.09.26
Context Engine
How Loophole's Context Engine Understands Your Codebase
06.29.26
AI Extensions
Introducing Loophole AI Extensions
06.08.26
Building the future
Estimating the Productivity of an AI Software Engineer
06.04.26
FAQ
Questions, answered.
No catch. Loophole is open-source under an MIT-style licence — you can read every line of code on GitHub. We don't charge for the editor. The only costs you'll ever pay are directly to your chosen AI provider for API usage, and those bills go straight to them, not through us.
No. Your API keys are stored locally on your machine in your own config file and are never sent to Loophole's servers — because Loophole doesn't have servers. Requests go directly from your editor to your provider's endpoint. We never see your keys, your prompts, or your code.
It depends entirely on which model you pick and how much you use it. Light everyday work on a model like Claude Haiku or Gemini Flash costs pennies per day. Heavy agentic sessions with frontier models can be a few dollars per session. You can also run entirely free using local models via Ollama or vLLM at zero cost per token.
Yes. Loophole supports any OpenAI-compatible endpoint, which means Ollama, vLLM, LM Studio, and any other local inference server work out of the box. You get the full editor experience — inline diffs, Plan mode, agentic runs — at zero token cost and complete air-gap privacy.
Currently, nothing. Both are the same open-source build downloaded from GitHub. The distinction exists to prepare for the Team plan launch, which will add shared infrastructure on top of the existing free core. Every feature you see today is in both and always will be.
We're actively building it but haven't set a date or price yet. Join the waitlist on the GitHub Discussions page and you'll be the first to know. We're committed to keeping it affordable — our philosophy is that tooling shouldn't eat your AI budget.
Only to the AI provider you explicitly configure. Your code is sent to whichever endpoint you've pointed Loophole at — OpenAI, Anthropic, your own Ollama instance, or anyone else. Nothing is sent to Loophole. The editor has no telemetry, no analytics, and no call-home to our infrastructure.
Absolutely. Loophole is built on Void, itself a fork of VS Code, and the entire codebase is on GitHub. PRs, issues, and forks are all welcome. If you build something interesting on top of Loophole, share it in Discussions — we actively spotlight community work.
EOF