🛡 Freeware Release v1.1.6

Your AI. Your Computer.
Owned Permanently.

The Sovereign Choice in local intelligence. Built for professional workflows where security and data privacy are not optional. No cloud, no subscription, no tracking.

⬇ Download LexSort v1.1.6 (Freeware)

Apache 2.0 Open Source · For macOS, Windows, and Linux

See It In Action

Real screenshots from LexSort Vera v1.1.6. No demos, no mock-ups — this is the actual app.

LexSort Vera VERA Chat — Private AI assistant running locally on your machine

VERA Chat — your fully private AI assistant. Runs 100% on-device, no account required.

LexSort Vera model selection — automatic hardware detection with Llama, Qwen, Mistral support

Automatic hardware detection selects the best local model for your machine — Llama, Qwen, Mistral and more.

LexSort Vera Quick Organizer module — AI-powered calendar and task management

Quick Organizer module — AI-powered task and calendar management, fully offline.

The Foundation of Trust

Why we build local-first and share the source code with the world.

🔒

100% Local

LexSort runs entirely on your device. Every token is computed locally. No remote servers, no cloud storage, and zero data sharing with third parties.

🔎

Audit-Ready

Verify our privacy claims. Our codebase is open source under Apache 2.0. Audit the network configurations, system calls, and code directly on GitHub.

⚡

Instant Boot

Built for raw performance. By bypassing remote servers, your queries initiate and stream instantly. No subscription queues, no latency spikes, no waiting.

The Product Roadmap

We believe in delivering a robust, complete free experience, while designing modules for upcoming power tiers.

LexSort Vera Freeware (v1.1.6)

Complete Workspace

An unrestricted, private desktop environment with no usage tracking.

✓ Hardware Auto-Detection: Adjusts LLM size based on your RAM automatically.
✓ Local Inference Engine: Executes locally using the Ollama HTTP API.
✓ RAM-Only Memory: Conversations live in volatile memory and vanish on close.
✓ Zero Telemetry: No crash reports, usage counters, or external calls.
✓ Unrestricted Queries: No token limits or speed throttling.

LexSort Vera Pro Roadmap

Enterprise Extensions

Advanced background agents and integration modules scheduled for Q4 2026.

⏳ Auto Emailer Module: Automatic, secure drafted replies for local inbox clients (Tested & Ready).
⏳ Guardian Watch: Background process to securely monitor local folders and document folders.
⏳ LexSort-GO Mobile: Companion mobile application with peer-to-peer sync.
⏳ iMessage & Email Ingestion: Connect local notifications to summarize context.
⏳ Encrypted Database: Optional local chat archives, secured with zero-knowledge keys.

Real Hardware. Real Speed.

Every number below was measured live on an Apple M1 Pro 16GB Mac — no synthetic tests, no marketing estimates. This is what you actually get.

🏆 Fastest

Llama 3.2 3B

45.5 tok/sec

2.0 GB download · Works on 4GB+ RAM

Instant, real-time responses. Best for everyday tasks, Q&A, and quick drafts.

⚡ Fast

Mistral 7B

33.9 tok/sec

4.1 GB download · Works on 8GB+ RAM

Excellent writing, reasoning, and general assistant tasks. The sweet spot for most users.

🎯 Balanced

Qwen 2.5 Coder 7B

24.3 tok/sec

4.7 GB download · 16GB+ RAM

High quality code generation and technical reasoning. Ideal for developers on 16GB Apple Silicon.

💎 Quality

Qwen 2.5 Coder 14B

12.4 tok/sec

9.0 GB download · 16GB+ RAM

Maximum local AI quality. Best for deep code review, architecture planning, and complex problem solving.

Tested on Apple M1 Pro, 16 GB unified memory, macOS 15 · June 2026 · All models run fully offline via Ollama

Your AI. Your Computer.Owned Permanently.