
Nexura models, persistent chats, owner controls, memory, research and a contained cybersecurity evaluation environment — brought together in one workspace.

When enabled, image bytes are sent directly from this device to Google Cloud Vision for Web Detection. The key is stored only in this browser's Nexura settings.
Nexura uses Google's Programmable Search Element in this browser and captures result titles, snippets and URLs for Research. No Google search API key is stored for this feature.
Install all models stores each public model package in persistent browser storage. Nexura keeps the selected model engine live for the active app session and reuses it for every message; a model is only reopened after a real page/browser restart, model switch, cache clear, or browser/GPU process loss.
Until paid plans launch, Nexura Free can use every public Nexura model: Nexura Lite, Core, Pro, Ultra and Max. The owner-only Nexura Testing Model is never included in any public plan. When plans go live, Free model access automatically changes to Lite + Core + Pro and Starter unlocks the full public model range.
The Testing Model stays owner-only and outside every subscription tier. No payment is taken while Starter says Coming Soon.
TXT, MD, JSON, code files and PDFs are added to chat context. Images are analyzed locally by Nexura Vision and can also be read for visible text.
The code unlocks the owner Testing Model for normal monitored chat. Unlocking it does not start Cyber Range Mode. Cyber Range starts only when ACTIVATE AUTONOMOUS RANGE is pressed.
Activation requires a previously installed local model. Research, Google image matching, shared-learning retrieval, backend audit RPCs and Cyber Range external-network tools are disabled while the range is active. The range exposes only simulated targets explicitly named by the exercise.