Mark VII

Enterprise-grade multi-provider AI chat platform with 45+ models, Firebase cloud configuration, and modern Material 3 design.

Features • Quick Start • Download • Documentation

Overview

Mark VII is a multi LLMs chat application for Android that provides unified access to 45+ state-of-the-art AI models from leading providers including Anthropic, OpenAI, Meta, Deepseek, Mistral, Google, and more through a single, elegant interface.

Key Capabilities

🤖 45+ AI Models - Gemini, GPT, Llama, Deepseek, Mistral, Groq-hosted models, and more
🔥 Cloud-First Architecture - Firebase-powered configuration management and real-time sync
⚡ High Performance - <100ms startup, connection pooling, optimized streaming
🔗 Triple API Support - OpenRouter (40+ models) + Direct Gemini API + Groq (ultra-fast inference)
📱 Modern Material 3 UI - Dynamic theming, smooth animations, haptic feedback
🌐 Multilingual TTS - Automatic language detection for 15+ languages
🔒 Enterprise Security - Firebase Authentication, encrypted storage, HTTPS only
🐍 DevOps Tools - Python CLI for bulk model management via CSV

Visuals

Usage across models from OpenRouter

Why Mark VII?

Flexibility: Switch between providers and models instantly without code changes
Reliability: Automatic error recovery, exception handling, offline capability
Performance: 24x faster than v1.x of Mark-VII, real-time streaming, optimized rendering
Developer-Friendly: Complete Firebase integration, comprehensive documentation, open source

Features

🤖 AI Model Access

Triple API Architecture:

OpenRouter Integration - Direct access to 100+ models with real-time catalog sync
Gemini API Integration - Native Google Gemini support with vision capabilities
Groq Integration - Ultra-fast inference API with all available Groq-hosted models
Seamless Switching - Toggle between all three APIs mid-conversation

Supported Companies:

Google - Gemini 2.0/2.5 Flash/Pro, Gemma 2/3
Anthropic - Claude 3.5 Sonnet, Opus, Haiku
OpenAI - GPT-4 Turbo, GPT-4o, GPT-3.5
Meta - Llama 3.1/3.3/4 (8B to 405B parameters)
Deepseek - Chat V3.1, R1, R1 Distill variants
Mistral AI - Full lineup from Small to Large
Groq - Llama 3 (8B/70B), Mixtral 8x7B, Gemma 2, Whisper (via Groq's ultra-fast inference)
Qwen (Alibaba) - Qwen2.5, Qwen3 Coder
xAI - Grok with vision support
30+ more - Cohere, AI21, Perplexity, and others

☁️ Firebase Cloud Integration

Remote Model Management - Add/remove models without app updates
Dynamic API Keys - Update credentials in real-time
Instant Sync - Changes reflect on next app restart
Secure Vault - API keys stored in Firebase, never in code
Offline Support - Cached configuration for offline operation
Exception Handling - Auto-detection and retry for :free suffix models
Smart Recovery - Automatic 404 error handling with model correction

💬 Advanced Chat Features

Real-Time Streaming - Server-sent events (SSE) for live responses
Multi-Modal Support - Text and image understanding (vision models)
Context Management - 6-message history for optimal performance
Session Persistence - Cloud-synced chat history with Google Sign-In
Brand Attribution - Clear provider identification (e.g., "Mark VII x Anthropic")
Smart Retry - Re-run prompts with different models
Stop Generation - Cancel responses instantly with red stop button
Voice I/O - Speech recognition input + multilingual text-to-speech
PDF Export - Professional formatting with syntax highlighting
Copy & Share - Easy text extraction and sharing

🎨 User Experience

Material 3 Design:

Dynamic theming (Light, Dark, System Default)
iOS-style color palettes with smooth transitions
Theme-aware status bar and navigation
No white flash on startup

Performance Optimizations:

<100ms startup time (24x faster than v1.x)
Connection pooling for API requests
Memoized rendering and lazy loading
Optimized scroll performance

Interaction Design:

Streaming cursor with haptic feedback
40dp touch targets for accessibility
Smooth Lottie animations
Auto-scroll to latest messages
Syntax highlighting for code blocks
Markdown support with inline formatting

Multilingual Support:

MLKit language detection (15+ languages)
Automatic TTS language switching
Support for: Chinese, Japanese, Korean, Spanish, French, German, Italian, Portuguese, Russian, Arabic, Hindi, English

🛠️ Management Tools

Efficiently manage 45+ AI models using CSV-based workflows.

Features:

Bulk Import/Export - Manage models in Excel/Google Sheets
Validation - Automatic format and field checking
Version Control - Track changes with Git
Interactive CLI - Menu-driven operations
Zero Downtime - Update models without app redeployment

Python CLI Usage:

cd update_models

# Import models from CSV (recommended)
python update_firebase_models.py --csv models.csv

# List current models in Firestore
python update_firebase_models.py --list

# Interactive mode
python update_firebase_models.py

# Export models to CSV
python update_firebase_models.py --export my_models.csv

CSV Format:

apiModel,displayName,isAvailable,order
google/gemini-2.0-flash-exp,Gemini 2.0 Flash,TRUE,1
deepseek/deepseek-chat-v3.1,Deepseek Chat V3.1,TRUE,2
anthropic/claude-3-5-sonnet-20241022,Claude 3.5 Sonnet,TRUE,3

Benefits:

Edit 50+ models in Excel/Google Sheets
Bulk enable/disable models
Easy reordering with sort priority
Version control friendly
Automatic validation

Quick Start

Prerequisites

Android Studio Ladybug or newer
JDK 17+
Firebase account (free tier sufficient)
OpenRouter API key (free tier available at openrouter.ai/keys)
Optional: Google Gemini API key for direct Gemini access
Optional: Groq API key for Groq inference (console.groq.com/keys)

1. Clone Repository

git clone https://github.com/daemon-001/Mark-VII.git
cd Mark-VII

2. Firebase Setup (~5 minutes)

Create Firebase Project

Go to Firebase Console
Click "Add project" → Enter name: Mark-VII
Disable Google Analytics (optional) → "Create project"

Add Android App

Click Android icon in Project Overview
Enter package name: com.daemon.markvii
Download google-services.json → Place in Mark-VII/app/ directory

Configure Firestore Database

Navigate to Firestore Database → "Create database"
Select "Start in test mode" → Choose region → "Enable"

Add Configuration Data

Option A: Python Script (Recommended)

cd update_models
pip install firebase-admin

# Download service account key:
# Firebase Console → Project Settings → Service Accounts
# → Generate New Private Key → Save as mark-vii-firebase-service-account-key.json

# Import 45+ models from CSV
python update_firebase_models.py --csv models.csv

Option B: Manual Setup

In Firestore, create collection: app_config
Create document: models
Add field: list (type: array) with model objects:

{
  "id": "anthropic/claude-3.5-sonnet",
  "name": "Claude 3.5 Sonnet",
  "provider": "anthropic",
  "isEnabled": true,
  "sortOrder": 1,
  "apiType": "openrouter"
}

Create document: api_keys
Add fields:
- openrouterApiKey (string) - Get from OpenRouter
- geminiApiKey (string, optional) - Get from Google AI Studio
- groqApiKey (string, optional) - Get from Groq Console

3. Build & Run

# In Android Studio:
# 1. File → Open → Select Mark-VII folder
# 2. File → Sync Project with Gradle Files
# 3. Build → Make Project (Ctrl+F9)
# 4. Run → Run 'app' (Shift+F10)

First Launch:

Grant internet permissions
Sign in with Google (creates cloud-synced chat sessions)
Select a model and start chatting!

Installation

End Users

Android Smartphone:

Download APK from Releases
Enable "Install from Unknown Sources" in Settings → Security
Open APK file → Install → Open Mark VII

Android Emulator (PC):

Download APK from Releases
Drag APK into emulator window or use APK installer
Launch Mark VII from app drawer

Developers

See Quick Start above for source-based setup.

Troubleshooting

"Firebase not configured" in app

Check Firestore structure: app_config/models document must have list field (not models)
Verify app_config/api_keys document has openrouterApiKey field
Ensure google-services.json is in app/ folder
Check exp_models collection for exception models

Groq API Issues

Ensure groqApiKey field is present in app_config/api_keys Firestore document
Or add your personal key in Settings → Groq API Key → Verify → Enable
Get a free key from console.groq.com/keys

HTTP 401 Error (Unauthorized)

Invalid API key
Get new key from OpenRouter
Update in Firebase: app_config/api_keys/openrouterApiKey

HTTP 404 Error (Model Not Found)

Model may require ":free" suffix
App automatically adds failing models to exp_models collection
Retry the request after error - it will use the correct format
Check model exists on OpenRouter Models

HTTP 429 Error (Rate Limit)

Too many requests
Wait a few seconds and retry
Consider upgrading OpenRouter plan

HTTP 502/503 Error (Server Issues)

OpenRouter or model provider temporarily down
Try a different model
Wait and retry later

App Crashing

Clear app data: Settings → Apps → Mark VII → Clear Data
Reinstall the app
Check for updates

Build Errors

# Clean and rebuild
./gradlew clean
./gradlew assembleDebug

Architecture

Overview

Mark VII uses a cloud-first architecture combining Firebase's real-time configuration management with triple AI API access (OpenRouter + Gemini + Groq), enabling zero-downtime model updates and enterprise-grade reliability.

┌─────────────────────────────────────────────────────────────┐
│                      Mark VII Android App                   │
├─────────────────────────────────────────────────────────────┤
│  UI Layer (Jetpack Compose + Material 3)                    │
│  ├─ MainActivity.kt         - Main UI orchestration         │
│  ├─ SettingsScreen.kt       - Theme & account management    │
│  └─ DrawerContent.kt        - Navigation & session list     │
├─────────────────────────────────────────────────────────────┤
│  ViewModel Layer (MVVM)                                     │
│  └─ ChatViewModel.kt        - State management + logic      │
├─────────────────────────────────────────────────────────────┤
│  Data Layer                                                 │
│  ├─ ChatData.kt             - OpenRouter API orchestration  │
│  ├─ GeminiClient.kt         - Direct Gemini API client      │
│  ├─ FirebaseConfigManager   - Remote model configuration    │
│  ├─ ChatHistoryManager      - Session persistence (Cloud)   │
│  ├─ AuthManager             - Google Sign-In + tokens       │
│  └─ ThemePreferences        - Local theme storage           │
├─────────────────────────────────────────────────────────────┤
│  Network Layer                                              │
│  ├─ Retrofit + OkHttp       - HTTP client with pooling      │
│  ├─ SSE EventSource         - Streaming response parser     │
│  └─ Connection Pool         - Persistent connections        │
└─────────────────────────────────────────────────────────────┘
              ↓                              ↓
    ┌───────────────────┐        ┌──────────────────────────┐
    │ Firebase Services │        │   AI API Providers       │
    ├───────────────────┤        ├──────────────────────────┤
    │ Firestore         │        │ OpenRouter (100+ models)  │
    │ Authentication    │        │ Direct Gemini API         │
    │ Analytics         │        │ Groq (ultra-fast)         │
    └───────────────────┘        └──────────────────────────┘

Tech Stack

Frontend:

Kotlin with Jetpack Compose for declarative UI
Material 3 Design System with dynamic theming
Lottie for animations, Markdown with syntax highlighting
StateFlow for reactive state management

Architecture Pattern:

MVVM (Model-View-ViewModel) with repository pattern
Coroutines + Flow for asynchronous operations
Dependency injection via constructor parameters

Backend Integration:

Firebase Firestore: Cloud configuration + chat history
Firebase Authentication: Google Sign-In with OAuth
Firebase Analytics: Usage tracking (optional)

Networking:

Retrofit 2.11.0 with OkHttp 4.12.0
Server-Sent Events (SSE) for streaming responses
Connection pooling for <100ms startup time

AI Providers:

OpenRouter API: 100+ models with unified interface
Direct Gemini API: Native Google integration with vision
Groq API: OpenAI-compatible, ultra-fast inference (Llama 3, Mixtral, Gemma)
MLKit Language ID: Automatic TTS language detection

Tools:

Python 3.11+ with Firebase Admin SDK
CSV-based model management workflow
Git for version control

Project Structure

Mark-VII/
├── app/
│   ├── google-services.json              # Firebase config (download from console)
│   └── src/main/java/com/daemon/markvii/
│       ├── MainActivity.kt                # App entry + Firebase initialization
│       ├── ChatViewModel.kt               # MVVM state management
│       ├── data/
│       │   ├── Chat.kt                    # Data models (Message, ChatRequest, etc.)
│       │   ├── ChatData.kt                # API orchestration (OpenRouter + Groq)
│       │   ├── GeminiClient.kt            # Direct Gemini API client
│       │   ├── OpenRouterApi.kt           # OpenRouter Retrofit interface + client
│       │   ├── GroqApi.kt                 # Groq Retrofit interface + client
│       │   ├── FirebaseConfig.kt          # Model configuration models
│       │   ├── FirebaseConfigManager.kt   # Firestore config operations
│       │   ├── AuthManager.kt             # Google Sign-In + token management
│       │   ├── ChatHistoryManager.kt      # Cloud chat session storage
│       │   ├── ThemePreferences.kt        # Local theme persistence
│       │   ├── UserApiPreferences.kt      # User API key storage (Gemini/OR/Groq)
│       │   └── Keys.kt                    # App metadata + build info
│       ├── ui/theme/
│       │   ├── Theme.kt                   # Material 3 theme + AppColors
│       │   └── Color.kt                   # Theme color definitions
│       ├── SettingsScreen.kt              # Theme selector + account UI
│       ├── DrawerContent.kt               # Navigation + session management
│       └── utils/
│           └── PdfGenerator.kt            # PDF export with syntax highlighting
├── update_models/
│   ├── update_firebase_models.py          # Model management CLI
│   ├── models.csv                         # Pre-configured 49 models
│   └── mark-vii-firebase-service-account-key.json  # Service account (download)
├── CHANGELOG.md                           # Git commit-based version history
├── FIREBASE_SETUP.md                      # Detailed Firebase setup guide
└── README.md                              # Project documentation

Usage Examples

Basic Chat Workflow

Launch - Open Mark VII (cold start <100ms)
Sign In - Tap profile icon → "Sign in with Google" (optional, enables cloud sync)
Select Model - Tap dropdown → Choose model (e.g., "Claude 3.5 Sonnet")
Input - Type message or tap microphone icon for voice input
Send - Tap send button (▲) → Get real-time streaming response
Attribution - See response header "Mark VII x Anthropic"
Stop - Tap red stop button (■) to cancel streaming anytime

Switching Between API Providers

Open model dropdown
Select different provider mid-conversation to compare responses
Chat history preserves which API was used per message

Theme Customization

Open drawer (swipe right or tap menu) → Tap Settings icon (⚙️)
Under "Appearance", tap Theme selector
Choose theme:
- System Default - Follows device settings (auto light/dark)
- Light - iOS-inspired bright theme with gentle shadows
- Dark - Eye-friendly with OLED-optimized blacks
Theme applies instantly without restart, status bar updates automatically

Managing Chat Sessions

Open navigation drawer (swipe right or hamburger menu)
View all sessions with preview of last message
Switch - Tap session to load conversation
Rename - Long-press → "Rename" → Enter new name
Delete - Long-press → "Delete" → Confirm
New Chat - Tap "+" button → Fresh session created
Cloud Sync - All sessions backup to Firebase when signed in

Image Understanding (Vision Models)

Tap plus icon (+) in input field
Select image from gallery or take photo
Type question about image (e.g., "What's in this image?")
Works with vision-capable models: Gemini 2.0 Flash, Claude 3.5, GPT-4o
Image preview shows pin indicator when attached
Send to get multi-modal analysis

Multilingual Text-to-Speech

Receive AI response in any language (Chinese, Spanish, Arabic, etc.)
Tap speaker icon on message
MLKit automatically detects language
TTS reads response in correct language/accent
Supports 15+ languages: Chinese (Mandarin), Japanese, Korean, Spanish, French, German, Italian, Portuguese, Russian, Arabic, Hindi, English, Dutch, Polish, Turkish

PDF Export

Complete conversation with AI model
Tap 3-dot menu → "Export to PDF"
PDF generates with:
- Syntax-highlighted code blocks
- Formatted Markdown rendering
- Message timestamps
- Model attribution (e.g., "Mark VII x Anthropic")
Share via any app or save to storage
Professional formatting for documentation/reports

Retry with Different Models

Send message to Model A (e.g., Claude 3.5 Sonnet)
Get response → Not satisfied with output
Tap model dropdown → Switch to Model B (e.g., GPT-4o)
Tap "Retry Last Prompt" button
Get alternative response from different AI
Compare responses side-by-side in chat history

Voice Input

Tap microphone icon in input field
Grant microphone permission (first time)
Speak your question naturally
Speech-to-text transcription appears automatically
Edit if needed, then send
Works in all languages supported by Android Speech Recognition

Performance

Current Metrics (v3.3.0)

Startup Performance:

Cold Start: <100ms (24x faster than v1.x)
Model Loading: <50ms (cached configuration)
Theme Application: <10ms (instant visual feedback)
Firebase Init: Asynchronous, non-blocking

Runtime Optimization:

Streaming Latency: <500ms to first token
Rendering: 60 FPS with memoized composables
Memory: <50MB baseline, <150MB during streaming
Network: Connection pooling reduces request overhead by 70%
TTS Language Detection: <100ms with MLKit on-device processing

Note: Times vary based on network, server load, and prompt complexity.

Version Comparison

Metric	v1.x (Gemini)	v2.x (OpenRouter)	v3.0+ (Dual API)	v3.3 (Triple API)
Startup Time	~2.65s	~110ms	<100ms	<100ms
Models Available	5-10	100+	100+ (dual APIs)	100++ (three APIs)
API Providers	1 (Google)	Multiple	OpenRouter + Gemini	+ Groq
Configuration	Hardcoded	Cloud-based	Cloud + real-time	Instant updates
Streaming	No	Yes (SSE)	Yes (both APIs)	Yes (all APIs)
Error Handling	Basic	Comprehensive	Auto-retry + 404 fix	Resilient
TTS Languages	1-2	1-2	15+ (MLKit)	15+ (MLKit)
Theme Support	Basic	Light/Dark	L/D/System + iOS-style	Polished
Offline Support	No	Limited	Cached config	Works offline
Stop Generation	No	Yes	Yes + haptics	User control

Contributing

We welcome contributions from the community! Whether you're fixing bugs, adding features, or improving documentation, your help makes Mark VII better.

Reporting Issues:

Open an issue
Include: Android version, app version, steps to reproduce
Add screenshots if helpful

Feature Requests:

Open a Feature Request with tag enhancement
Describe the feature and its use case
Provide mockups/examples if relevant
Discuss with maintainers before implementing

Code Contributions:

Fork the repository
Create feature branch: git checkout -b feature/your-feature-name
Make changes following Kotlin code style guidelines
Test thoroughly on multiple devices/Android versions
Commit with clear messages: git commit -m "Add: Feature description"
Push to your fork: git push origin feature/your-feature-name
Open Pull Request with detailed description

Dependencies

Core Libraries

// Firebase (BOM manages versions)
firebase-bom:33.7.0
firebase-firestore-ktx          // Cloud configuration + chat history
firebase-analytics-ktx          // Usage analytics
firebase-auth-ktx               // Google Sign-In authentication

// Google Services
play-services-auth:21.3.0       // Google Sign-In UI
mlkit-language-id:17.0.6        // Multilingual TTS detection

// Networking
retrofit:2.11.0                 // Type-safe HTTP client
okhttp:4.12.0                   // Connection pooling + SSE
gson:2.10.1                     // JSON serialization

// PDF Generation
itext7-core:7.2.5               // PDF document creation
html2pdf:4.0.5                  // HTML to PDF conversion

// Markdown & Syntax Highlighting
compose-markdown:0.5.4          // Rich text rendering
code-highlight:2.0.0            // Syntax highlighting for code blocks

// UI & Animation
androidx.compose.bom:2024.12.01 // Jetpack Compose
androidx.material3:*            // Material 3 components
lottie-compose:6.0.0            // Lottie animations
coil-compose:2.4.0              // Image loading

Minimum Requirements

Android SDK: 24 (Android 7.0 Nougat) or higher
Target SDK: 35 (Android 15)
Kotlin: 2.1.0
Gradle: 8.7
JDK: 17+

Security & Privacy

Data Protection

Encrypted Storage: API keys secured in Firebase, never in source code or APK
HTTPS Only: All API communication uses TLS 1.3 encryption
OAuth 2.0: Google Sign-In with secure token management
Local Storage: Chat history cached locally, synced to Firebase when signed in
No Tracking: Zero third-party analytics or ad networks
Firestore Rules: Read/write access restricted to authenticated users

Configuration Security

google-services.json excluded from Git via .gitignore
mark-vii-firebase-service-account-key.json excluded (Python tool only)
Firebase project rules restrict access to authenticated users
API keys rotatable without app updates (cloud configuration)

User Control

Sign Out Anytime - Settings → Sign Out → Clears local cache
Delete Chat Sessions - Long-press session → Delete (removes from cloud)
Theme Preferences - Stored locally, never synced
Optional Cloud Sync - Chat history syncs only when signed in with Google
Offline Mode - Works with cached configuration when no internet

Best Practices

Never commit google-services.json or service account keys to public repos
Rotate Firebase API keys periodically in Firebase Console
Use Firebase Authentication rules to restrict Firestore access
Enable Firebase App Check for production builds to prevent API abuse

Support & Contact

Get Help

Bug Reports:

GitHub Issues
Include: Android version, app version, steps to reproduce

Feature Requests:

Feature Request Form
Describe use case and expected behavior

Documentation:

CHANGELOG.md - Version history with git timestamps
FIREBASE_SETUP.md - Detailed Firebase configuration
README.md - Comprehensive project documentation

Email Support:

Developer: nitesh.kumar4work@gmail.com
Response time: 1-12 hours

Connect

Developer:

Name: Nitesh Kumar
GitHub: @daemon-001
LinkedIn: @daemon001

Project:

Repository: Mark-VII
Stars: Give us a ⭐ if you find this useful!
Forks: Welcome - see Contributing section

Every contribution, no matter how small, helps make Mark VII better for everyone!

Built with ❤️ by Nitesh

⬆ Back to Top ⬆

Home • Download • Report Bug • Request Feature

Enjoy your advanced multi-LLMs AI chatbot experience!

Name		Name	Last commit message	Last commit date
Latest commit History 83 Commits
.idea		.idea
app		app
docs/visuals_3.1.0		docs/visuals_3.1.0
gradle/wrapper		gradle/wrapper
update_models		update_models
.gitignore		.gitignore
CHANGELOG.md		CHANGELOG.md
FIREBASE_SETUP.md		FIREBASE_SETUP.md
LICENSE		LICENSE
README.md		README.md
build.gradle.kts		build.gradle.kts
gradle.properties		gradle.properties
gradlew		gradlew
gradlew.bat		gradlew.bat
settings.gradle.kts		settings.gradle.kts

Folders and files

Latest commit

History

Repository files navigation

Mark VII

Overview

Key Capabilities

Visuals

Usage across models from OpenRouter

Why Mark VII?

Features

🤖 AI Model Access

☁️ Firebase Cloud Integration

💬 Advanced Chat Features

🎨 User Experience

🛠️ Management Tools

Quick Start

Prerequisites

1. Clone Repository

2. Firebase Setup (~5 minutes)

Create Firebase Project

Add Android App

Configure Firestore Database

Add Configuration Data

3. Build & Run

Installation

End Users

Developers

Troubleshooting

"Firebase not configured" in app

Groq API Issues

HTTP 401 Error (Unauthorized)

HTTP 404 Error (Model Not Found)

HTTP 429 Error (Rate Limit)

HTTP 502/503 Error (Server Issues)

App Crashing

Build Errors

Architecture

Overview

Tech Stack

Project Structure

Usage Examples

Basic Chat Workflow

Switching Between API Providers

Theme Customization

Managing Chat Sessions

Image Understanding (Vision Models)

Multilingual Text-to-Speech

PDF Export

Retry with Different Models

Voice Input

Performance

Current Metrics (v3.3.0)

Version Comparison

Contributing

Dependencies

Core Libraries

Minimum Requirements

Security & Privacy

Data Protection

Configuration Security

User Control

Best Practices

Support & Contact

Get Help

Connect

About

Resources

License

Uh oh!

Stars

Watchers

Forks

Releases 5

Packages 0

Uh oh!

Contributors

Uh oh!

Languages

Packages