Ogun is our purpose-built local model for offensive security — the brain behind Apophis, trained on real offensive work and real engagement data. It runs entirely in your environment, so nothing leaves your boundary at runtime.
Autonomous offensive security means feeding an engagement's most sensitive detail to a model. Ogun is how that model stays yours.
Served entirely local through Ollama. Your targets, tool output, and findings never leave your environment at runtime — no cloud, no third-party inference.
Tuned on offensive-security knowledge and real Apophis engagement data against real targets — the way an operator actually works, not generic web text.
Purpose-built to drive Apophis's tool-calling loop and produce its exact CVSS-mapped report format. A drop-in local brain, not a general chatbot bent to the task.
Quantized to run on a single box, with a context window you size to the engagement. Built to serve one team's agents fast, not to top a leaderboard.
Ogun already runs as the model behind Apophis in our own environment. It's early — here's the honest state of it.
Ogun is the Yoruba orisha of iron, war, and craft — the force a smith turns to at the forge, and the patron of everyone who works metal into tools and weapons. The right namesake for the model we forge in-house to arm our own agents.
Ogun is in early development. Tell us about your environment and we'll bring you in as it matures.