LocalMind AI Offline
Loading...
LocalMind AI
WORKSPACE/Painel
BETA PREVIEW☼

ON-DEVICE INTELLIGENCE

Your AI space

All set to run locally — once the Android engine is connected.

+New conversation
!
Android bridge not connected

This web preview does not run GGUF or host an API. The llama.cpp engine, Android service, and native storage need to be integrated into the Capacitor project.

×
AI ENGINEOff
Not connectedRequires native integration
DEFAULT MODEL◈
NoneSelect a GGUF model
API SERVERStopped
UnavailableAndroid service not installed
ADDED MODELS▱
0Metadata in this preview

Overview

LIVE
⌁Device resources
Details
RAM memoryIndisponível
CPU usageIndisponível
Metrics available after native integration
✦Get started here
01Add a GGUF modelOrganize your models on this device→02Configure API accessSee requirements to connect a PC→

PRIVATE CHAT

Chat

Chat with a model running on your own device.

◈NoneMotor não conectadoOFFLINE⌄
—CPU
—RAM
—TOK/S
Clear conversation+ New conversation
✦

A conversation, just for you.

Add a GGUF model and connect the Android engine to start chatting.

Organize my ideasExplain a concept

AVAILABLE MODELS

Choose a model

×

Select an imported model to use in the conversation.

No models imported yet. Add a GGUF file in the Models section.

In the web preview, models appear as metadata only and cannot be loaded.

LOCAL LIBRARY

Models

Manage the GGUF files you want to use with LocalMind.

i
Native import required

The preview can read the name and size of the chosen file, but cannot copy it to persistent storage or load it into llama.cpp. On Android, connect the document picker and the native bridge.

AllDefault
0 modelos

No model is selected as default.

◈

Your library starts here

Add .gguf files to keep your collection organized. Inference requires native llama.cpp integration.

+ Import first model

Performance

CPU—Native metric
RAM—Native metric
Total RAM—Native metric
Tokens/s—Native metric
⚙Inference settings
Saved in this app
ThreadsContext sizeOutput token limitMemory budget (MB)

The memory target is not a guarantee: consumption varies by model, context, caches, and operating system.

Save settings

LOCAL NETWORK ACCESS

API Server

Use your phone as a local endpoint for apps on the same Wi-Fi.

Parado
!
Native server unavailable in this preview

Web preview does not open ports. In the compiled Android app, the local server works while the app is active, on the same network, and with a loaded model.

×
⌘Server control
Parado
Port
Iniciar servidor

Use a port between 1024 and 65535.

Base URLhttp://IP_DO_CELULAR:8080/v1Copy
⌑Access keys
+Generate key

Keys created here are stored in this app. The preview does not authenticate requests without the native service.

No keys created.

{ }PC Example
Copy
curl http://IP_DO_CELULAR:8080/v1/chat/completions \
  -H "Authorization: Bearer SUA_CHAVE" \
  -H "Content-Type: application/json" \
  -d '{"model":"local","messages":[{"role":"user","content":"Olá!"}]}'

On Android, use the displayed address, a valid key, and a loaded model; phone and computer must be on the same local network.

Export project files

SFEU · Patented by Eric Almeida Montanario

PROJECT EXPORT

Download files as ZIP

×

Enter the password to continue.

PasswordContinue

Enter the password and generate a ZIP with the project files. If any file cannot be read automatically, select the complete project folder.

The password is a simple confirmation; the ZIP file will not be encrypted.

BackGenerate project ZIP

Draw on screen Pen Eraser CPU +10%Clear

Draw with your finger or mouse. Tap close to go back to using the screen.

Draw