Pulse is the dashboard. NYMPH AI is the 4.5B-parameter model that powers it — installed locally, no tokens metered, no data leaving your device.
3 days free. No credit card. Works in any modern browser.
Chat, browse your activity, run the agent on your machine, organize references, generate images, monitor spend, message your team. Five hero surfaces and a few more reachable from the command palette.
A proprietary 4.5B model wrapped in a cognitive runtime that handles tool calling, retrieval, output validation and persistent memory. The hard part — the runtime — is what makes a small model behave like a much larger one.
Every part of Pulse runs against the same local model. No context-switching, no copy-paste, no juggling tabs.
Streaming conversations with NYMPH AI. Multi-file attachments, drag-and-drop, slash commands, inline search. Switch to Claude or GPT-4 for a second opinion when you want it.
Goal-directed loop with tool execution. Calculator, web search, file ops — the model decides which tool to call and shows you the trace.
Projects with persistent context. Drop in docs, conversations imported from ChatGPT/Claude, or URLs. Every chat in the project sees them.
Server health, agent metrics, GPU temperature, API spend across providers. Activity Log so you remember what you did and where you left it.
Image generation built in. ENSO style models for art, DALL-E for fallback. Saved to your private gallery.
Real-time messaging with teammates and AI personas. Drag-drop file sharing, unread badges, end-to-end via Supabase.
Because the model runs on your hardware, there are no inference costs to pass on. Pay once, use forever.
Try Pulse without a card.
NYMPH AI on your machine.
Pay once. Yours forever.
The web app runs in any modern browser. The local model takes one terminal command to install. Both free for 3 days.