Featured
Featured
The latest from Ophraxx — launch news, feature releases, engineering notes, and development stories.
Launch
Ophraxx AI Discord Bot is now in beta
Ophraxx AI has entered public beta on Discord. Servers can now add the bot, run the interactive setup command, and start chatting with Ophraxx Lite • OX-L1 — complete with spam protection, usage limits, mod log integration, and a full five-layer safety pipeline.
Feature
Five personality modes now available for server admins
Server admins can now configure how Ophraxx AI communicates through five distinct personality modes — Default, Professional, Friendly, Tutor, and Concise. Each mode changes tone and style while leaving safety rules completely untouched.
Safety Research
Designing the S1–S9 threat model: how we categorized safety risks
A research note on the decisions behind Ophraxx AI's nine-category threat model — why each category exists, how the safe-topic allowlist prevents over-refusal, and the reasoning behind the S8 and S9 categories that cover jailbreaks and persona abuse.
More featured
Browse updates
Building Ophraxx AI's multi-layer safety stack
A look at the real safety architecture powering Ophraxx AI — from LLM-based input screening and nine-category threat detection to automated fact-checking, PII redaction, and the philosophy behind requiring explanatory refusals.
Ophraxx AI's first live data tool: real-time weather
Ophraxx AI can now answer weather questions with real, current data — not guesses. When a weather query is detected, the bot fetches live conditions and injects them directly into the AI's context before generating a response.
What's next: Ophraxx Lite • OX-B12 and OX-U45
Ophraxx Lite • OX-B12 and OX-U45 are the next two model tiers coming to the platform. Here's what each one is designed for, how the prompt classifier already prepares for them, and what the upgrade path looks like for server admins.
From Discord message to AI reply: inside the Ophraxx pipeline
A detailed walkthrough of everything that happens between a user sending a message in Discord and receiving an Ophraxx AI reply — spanning spam detection, rate limiting, model routing, live tool calls, conversation memory, and chunked output delivery.
Ophraxx Lite: how we built our model lineup and routing system
An inside look at the three-model Ophraxx Lite lineup — what each model is designed for, how the prompt classifier automatically routes queries between them, how the five personality modes work, and what's coming next.