Your Private AI.
On Your Silicon. Zero Cloud Lag.
Chat with state-of-the-art LLMs, analyze photos and documents, synthesize 4x HD art, study with interactive notebooks, and converse naturally in voice—all executed directly on your smartphone hardware with instant sub-20ms token latency. Zero mandatory account or sign-in needed.
// architectural advantages
Why On-Device AI Outperforms Cloud Services
Traditional AI apps turn your private data into server telemetry and lock you behind monthly cloud compute fees. ORUS runs natively on your phone silicon.
Air-Gapped Privacy
Your conversations, photos, and documents never leave your phone. What happens on your device stays on your device.
Instant Velocity (TTFT <20ms)
Eliminates network ping and API queues. Local NPU acceleration delivers typewriter token streaming in sub-20 milliseconds.
Complete Offline Autonomy
Fully operational on airplanes, remote expeditions, and underground transit. Zero mandatory account, cell data, or Wi-Fi required.
All-in-One Multimodal Suite
Smart chat, real-time voice, vision OCR, local PDF analyzer, 4x HD image studio, and knowledge notebooks in one unified app.
Transparent Tier Economy
Core Free on-device inference forever ($0 compute fees), with Pro ($4/mo) and Ultra ($8/mo) unlocking 32k/128k context, voice duplex, and Image Studio.
Super-Optional Encrypted Sync
Account connection and cloud backup are 100% optional. If you choose to enable sync via Supabase, it is end-to-end encrypted.
// interface walkthrough
Engineered for Pure Fluidity
Clean OLED black interface. Zero cloud dependency. What you see is what runs directly on your smartphone.
Typewriter Token Streaming & Rich Markdown
Fluid on-device inference with LaTeX math, syntax highlighted code blocks, message editing, and one-tap copy actions.
// 12 multimodal suites
A Complete On-Device AI Superapp
Twelve comprehensive modules engineered to run locally on your phone with zero server overhead.
Ultra-Fast Local Chat
Typewriter token streaming with sub-20ms latency, LaTeX math rendering, and multi-language syntax highlighting.
- •Sub-20ms Time to First Token
- •One-tap code copy & response regeneration
- •Context memory & multi-session management
Multimodal Vision & OCR
Attach photos or point your camera to extract text, decipher handwriting, tabulate receipts, and inspect diagrams.
- •On-device visual understanding
- •Zero photo uploads to external servers
- •Visual troubleshooting & document parsing
Document Reader & PDF Analyzer
Drop in contracts, textbooks, and reports for instant executive summaries, data extraction, and semantic Q&A.
- •High-throughput on-device PDF parsing
- •Deep clause and citation extraction
- •100% confidential business file review
Real-Time Offline Voice Mode
Hands-free voice assistant with real-time Voice Activity Detection (VAD) and natural neural speech synthesis.
- •Accurate offline speech-to-text dictation
- •Expressive natural voice packs
- •Ultra tier: Live duplex conversational loop
100% On-Device Image Studio
Synthesize artwork, photography, and digital illustrations directly on your phone using neural diffusion models.
- •Square, portrait (9:16) & widescreen (16:9) ratios
- •Granular CFG scale, sampling steps & seed control
- •Photorealistic and anime style presets
Neural HD Upscaler (4x AI Super-Res)
Quadruple image resolution into razor-sharp 2048px high-definition masterworks in under 5 seconds.
- •Hardware-accelerated neural enhancement
- •Modes for photography and anime art
- •Instant in-gallery replacement
Knowledge Notebooks & AI Quizzes
Organize research vaults for study topics and generate active-recall quizzes and flashcards automatically.
- •Structured notebooks with custom icons
- •One-tap active-recall quiz generator
- •Automated scoring & step-by-step explanations
Curated Model Hub & GGUF Import
Browse optimized open-weight LLMs with clear RAM requirement badges, or import custom GGUF models directly.
- •Quantized models optimized for ARM mobile silicon
- •Background download manager with resume
- •Direct Hugging Face & local file import
Plugin Studio & Tools
Augment your assistant with internal calculation engines, date/time awareness, and custom automated workflows.
- •Built-in system tools & math engine
- •Custom system prompt injection
- •Multi-step plugin chains
Optional Hybrid Cloud & BYOK
Bring your own API keys for Claude 3.5 Sonnet, GPT-4o, and Gemini 1.5 Pro with encrypted Android Keystore storage.
- •Direct encrypted client-to-API connection
- •Zero ORUS intermediary proxy or logging
- •Seamless fallback when phone RAM is low
OLED Dark Interface & Inks
Pure black #000000 interface engineered for minimum battery consumption with customizable accent ink palettes.
- •True black OLED battery savings
- •Personalized bubble colors and fonts
- •Fluid spring physics & tactile haptics
Data Sovereignty & Encrypted Sync
Complete local data ownership with one-tap JSON export and optional end-to-end encrypted multi-device backup.
- •One-tap JSON & Markdown chat export
- •Zero-trace instant wipe
- •Pro/Ultra: Encrypted Supabase multi-device sync
Predictable Plans. Zero Server Markup.
Because ORUS runs on your device silicon, you never pay for cloud GPU compute. Free on-device inference forever, with Pro ($4/mo) and Ultra ($8/mo) unlocking unconstrained hardware horsepower.
Core4,096 tokens
Essential on-device AI for privacy-conscious daily use.
Free forever
Included Capabilities:
- Unlimited local text inference (zero server costs)
- Up to 3 active downloaded models in local storage
- Standard 4k context window with CPU execution
- Standard offline voice dictation & 1 default TTS voice
- Single document PDF/text analysis
- 100% air-gapped privacy with zero data tracking
- No account or cloud connection required
- •On-device Image Generation Studio
- •Local RAG Knowledge Base & Voice Duplex
Pro32,768 tokens
Uncapped intelligence for creators, developers & knowledge workers.
$32 billed annually (in USD)
Included Capabilities:
- Unlimited local model downloads & storage
- Expanded 32k context window with Flash Attention
- 100% On-device Image Generation Studio (Diffusion)
- All offline natural voice packs & speech engines
- Full Plugin Studio (custom system prompts & plugins)
- Super-optional Encrypted Cloud Sync (connect only if you want)
- Multi-document PDF & Image Vision OCR analysis
- KV cache quantization (Q8/Q4) & accelerated inference
- •Neural 4x HD Upscaler & Live Voice Duplex
Ultra65k – 128k tokens
Peak performance for researchers, engineers & AI power users.
$64 billed annually (in USD)
Included Capabilities:
- Everything included in Pro
- Frontier 65,536 – 128,000 token context window
- Neural HD Upscaler (4x AI super-resolution enhancement)
- Maximum CPU acceleration with ARM KleidiAI & OpenCL
- Live Voice Duplex Conversation (real-time voice loop)
- On-Device Knowledge Base (Local RAG & Embeddings)
- Hybrid Cloud Fallback (Claude 3.5, GPT-4o, Gemini 1.5 Pro)
- Background batch processing & Home screen widgets
- Priority access to experimental inference engines
100% Optional Cloud Sync & Zero Mandatory Account Creation
Connecting an account or cloud sync is completely optional. If you prefer, keep everything 100% offline without signing in.
// device requirements & engine
Optimized for Modern Android Silicon
Engineered from bare metal up. ORUS orchestrates mobile CPUs, GPUs, and NPUs to run inference at maximum speed with minimal thermal draw.
Android 10.0+
64-bit ARMv8 / ARMv9 architecture with scoped storage sandboxing
4 GB RAM
Sufficient for compact 1B–3B text models and voice dictation
8 GB – 12 GB+ RAM
Optimal for 7B–9B LLMs, 100% on-device diffusion & 4x upscaling
NPU / GPU Compute
ARM KleidiAI, Qualcomm Hexagon NPU, Vulkan & OpenCL tuning
~1.5 GB – 4 GB
Per downloaded model file. Delete, pause, or swap models anytime
BYOK & Supabase
100% optional direct client connections to Claude, GPT-4o & Gemini
// target applications
Built for High-Stakes Workflow
Whether guarding corporate IP or working in air-gapped field environments, ORUS delivers private, real-time intelligence anywhere on earth.
Privacy Advocates & Legal Teams
Analyze sensitive contracts, patient records, and confidential notes without transmitting trade secrets or PII to third-party cloud data centers.
Engineers & Researchers
Debug code, generate algorithms, and execute math formulas with syntax highlighting, LaTeX rendering, and zero rate limit throttling.
Travelers & Field Operatives
Stay 100% productive on long-haul flights, underground subways, and remote expeditions without relying on cellular data or Wi-Fi.
Students & Lifelong Learners
Ingest textbooks and lecture slides to generate automated active-recall quizzes, study flashcards, and conceptual deep-dives in custom notebooks.
Coming Soon to Android
ORUS is currently in final performance optimization and testing. The application will be available for public download very soon. Prepare to experience true on-device multimodal intelligence on your smartphone with zero cloud dependencies.
// knowledge base
Frequently Asked Questions
Everything you need to know about on-device execution, subscription tiers, and privacy.