High-Privacy Local LLM Solutions
for On-Device Use

Try Pulsar
Pulsar is available on Linux, Windows and Mac

Pulsar :
Your New AI Assistant
for everyday use.
Ideal for personal and business use.

Pulsar, developed by AstraMind, is an innovative application designed to efficiently manage and run advanced language models directly on your personal device, completely eliminating concerns about sharing personal data. It enables you to create highly personalized, multidisciplinary AI assistants for conversation, all while maintaining full privacy and control over your data.

Pulsar is available on Linux, Windows and Mac
More

Create the best Personality
for your Assistants

Pulsar allows you to create custom personalities to chat with, simulating real and fictional characters. You can manually set all personality spectra or generate them automatically, adjust the settings in real time and have everything ready to use instantly.
Multiple Users

We support multiple users simultaneously on local systems.

Security

All information is stored locally on the device. You no longer have to worry about your privacy.

The Best Personalized
AI Assistant

Select from thousands of LoRAs and personalities to instantly customize your LLM, all directly within the app.

Intuitive UI

Our team has developed a User Interface designed to be intuitive and easy for anyone to use. AI has never been easier.

Quick and Easy Installation

We handle it all, optimizing your computing resources from the first installation for the best experience.

Cross-Platform

We support all types of systems. You can use Pulsar UI on Windows, Linux and Mac
(Android and iOS soon!)

Latest Blog posts

Auralis
Auralis
Unofficial implementation for the paper "Mixture-of-Depths"
Unofficial implementation for the paper "Mixture-of-Depths"
AstraMind: A new perspective for Generative AI
AstraMind: A new perspective for Generative AI
Exploring the Architectural Innovations of AstraQuasar-4B 32K
Exploring the Architectural Innovations of AstraQuasar-4B 32K
BitMat: Improving Ternary Matrix Multiplication with Triton
BitMat: Improving Ternary Matrix Multiplication with Triton