Mark-VII is a 100% free and open-source multi-provider AI client built natively for Android. Instead of being locked into a single ecosystem, Mark-VII acts as a unified interface to interact directly with the world's most powerful LLMs via their native APIs.

Built with Kotlin and Jetpack Compose under a clean MVVM architecture, it's designed to be fast, lightweight, and completely transparent about how your data is handled.

CORE ARCHITECTURE & FEATURES

Multi-Provider Integration Seamlessly switch between AI providers on the fly. We currently support:
• Gemini API: Native support for Gemini Pro and Flash models.
• Groq API: Ultra-low latency inference for open weights like Llama 3, Mixtral, and Gemma.
• OpenRouter API: A unified gateway to 100+ models including Claude 3.5 Sonnet, GPT-4o, and DeepSeek.

Bring Your Own Key (BYOK) Mark-VII is built privacy-first. While a default free-tier is provided for testing, power users can input their own API keys to bypass rate limits and utilize their own billing. Keys are strictly stored locally on your device via EncryptedSharedPreferences and are only transmitted directly to the respective provider endpoints.

Developer-Centric Rendering Because we use AI to write code, the UI is optimized for it:
• Full Markdown Support: Tables, lists, and inline formatting are natively rendered.
• Code Highlighting: Syntax highlighting for all major programming languages.
• One-Tap Copy: Instantly copy code blocks without struggling with text selection.

Real-Time SSE Streaming Responses aren't buffered. Mark-VII utilizes Server-Sent Events (SSE) to stream tokens directly to the UI in real-time, coupled with a 60fps Compose renderer and localized haptic feedback for a highly responsive feel.

Cloud Sync (Optional) If you want to keep your chat history synchronized across devices, you can authenticate via Google Sign-In (Firebase Auth). Your chat sessions are then securely synced to a Firestore database. Don't want to log in? The app works perfectly offline and in Guest Mode.

Hardware Integration
• Custom Speech-to-Text: Native microphone integration with a real-time RMS audio visualizer for hands-free prompting.
• ML-Kit Text-to-Speech: Auto-detects the language of the AI's response and dynamically switches TTS engines for accurate pronunciation.

Open Source & Community No ads. No hidden tracking. No premium subscriptions. Mark-VII is an open-source project built for the community.

Check out the source code, inspect the architecture, or contribute to the project directly on our repository!