Free security work, without refusals.
The Defense models are free abliterated models in free-pi, with a bundled security-audit skill. Sign in with GitHub and pick one in /model.
Opt-in · in /model
Authorized use onlyCyber Defense
Your security work should not stall because a model refused lawful work on your own systems.
Defense is an opt-in section of the /model picker for signed-in GitHub users. It lists models for security work: finding and fixing vulnerabilities, hardening, and auditing code and systems you own or are authorized to test. Defense models are abliterated: open-weight models with their refusal behavior removed, so they help with lawful security work that general models decline.
Defense models
- Abliterated Large Flash~20x usage
- Abliterated Large V2~60x usage
- DeepSeek V4.1 Flash Abliterated~20x usage
- Gemma 4 Uncensored~20x usage
- GLM 4.7 Flash Heretic~20x usage
- Qwen 3.6 Plus Uncensored~20x usage
- Uncensored 1.2~20x usage
Bundled with free-pi
- Bundled security-audit skillfree-pi ships the audit method and reference checklists from Cloudflare's open-source security-audit skill (MIT). free-pi runs it in a single agent, without sub-agents or an OS sandbox, so the agent follows the method itself and you validate each finding yourself.
- Reverse engineering with
/reverse-engineer-anythingAnalyze software you own or may lawfully test, such as native binaries, JS and Electron apps, APKs, .NET assemblies and websites.
For systems you own, employer systems you have permission to test, and in-scope bug bounty targets. Do not use it against anyone else's systems.