ModelPulse-CLI

Lightweight Terminal AI Model Real-time Monitoring & Intelligent Recommendation Engine

简体中文 | 繁體中文 | English

---

Introduction

ModelPulse-CLI is a zero-dependency, cross-platform terminal tool that helps developers monitor AI model API status, compare model performance, estimate costs, and get smart recommendations based on their specific use case.

Track 21 models across 7 major providers (OpenAI, Anthropic, Google, DeepSeek, Meta, Qwen, Mistral) with benchmark scores, pricing data, and intelligent scoring algorithms.

Core Features

• Model Registry - Browse 21 AI models with detailed specs, benchmarks, and pricing
• Smart Recommendation - Get personalized model suggestions for 7 task types (coding, reasoning, vision, etc.)
• API Monitoring - Real-time endpoint health checks with latency measurement
• Side-by-Side Comparison - Compare any models across all metrics
• Pricing Dashboard - Cost analysis with cheapest/most expensive detection
• Multi-Format Export - Export to JSON, CSV, or Markdown
• Historical Tracking - 90-day history of monitoring and recommendation data
• Zero Dependencies - Pure Python stdlib, no pip install needed
• Cross-Platform - Works on Linux, macOS, and Windows

Quick Start

Option 1: Run Directly (No Install)

# Clone the repository
git clone https://github.com/gitstq/ModelPulse-CLI.git
cd ModelPulse-CLI

# Run directly
PYTHONPATH=src python3 -m modelpulse.cli list

Option 2: Install via pip

git clone https://github.com/gitstq/ModelPulse-CLI.git
cd ModelPulse-CLI
pip install -e .

# Now use the command directly
modelpulse list

First Commands

# List all tracked models
modelpulse list

# List models sorted by price
modelpulse list --sort price

# Filter by provider
modelpulse list --provider openai

# Monitor API endpoint status
modelpulse monitor

# Get model recommendation for code generation
modelpulse recommend --task code_generation

# Compare models side by side
modelpulse compare "gpt-4o,claude-sonnet-4-20250514,gemini-2.5-pro"

# View pricing dashboard
modelpulse pricing

# Export to CSV
modelpulse export --format csv -o models.csv

# View available task types
modelpulse tasks

Detailed Usage Guide

modelpulse list - Model Registry

List all tracked AI models with detailed information.

Flag	Description
--provider, -p	Filter by provider name (e.g., openai, anthropic)
--category, -c	Filter by category (flagship, cost_effective, reasoning, open_source, specialized)
--sort, -s	Sort by: name, price, context, mmlu, math

modelpulse recommend - Smart Recommendation

Get AI model recommendations based on task type.

Flag	Description
--task, -t	Task type (required): general_chat, code_generation, reasoning, vision, long_context, cost_sensitive, function_calling
--top, -n	Number of recommendations (default: 5)
--api-key	Only recommend from providers with configured API keys

Available Task Types:

• general_chat - General conversation and Q&A
• code_generation - Code writing and software engineering
• reasoning - Complex mathematical and logical reasoning
• vision - Image understanding and visual analysis
• long_context - Processing long documents and conversations
• cost_sensitive - High-volume, cost-optimized usage
• function_calling - Agent workflows and tool use

modelpulse monitor - API Endpoint Monitoring

Check real-time availability and latency of provider API endpoints.

Flag	Description
--timeout, -t	Request timeout in seconds (default: 10)

modelpulse compare - Model Comparison

Compare models side by side across all metrics.

modelpulse compare "gpt-4o,claude-sonnet-4-20250514,gemini-2.5-pro"

modelpulse pricing - Pricing Dashboard

View and compare pricing across all models.

Flag	Description
--provider, -p	Filter by provider

modelpulse export - Data Export

Export model database to various formats.

Flag	Description
--format, -f	Output format: json, csv, markdown
--output, -o	Output file path

modelpulse config - Configuration

Manage API keys and preferences.

# Show current config
modelpulse config --show

# Set API key for a provider
modelpulse config --set-key openai=sk-your-key-here

modelpulse history - Historical Data

View past monitoring and recommendation records.

Flag	Description
--type, -t	Filter by type: monitor, recommend
--limit, -n	Number of records to show

Design Philosophy

1. Zero Dependencies - Built entirely on Python stdlib. No pip install required to run.
2. Offline-First - All model data is embedded. Works without internet for listing/comparison/recommendation.
3. Extensible - Easy to add new models, providers, and task profiles by editing a single data structure.
4. Developer-Friendly - Clean output formatting, color support, and structured data export.
5. Cross-Platform - Tested on Linux, macOS, and Windows with no platform-specific dependencies.

Roadmap

• Web dashboard mode (HTML report generation)
• Auto-update model database from provider APIs
• Token usage tracking and budget alerts
• Plugin system for custom scoring algorithms
• Interactive TUI mode with keyboard navigation
• Model fine-tuning cost calculator
• Multi-language model name support (CN/JP/KR)
• CI/CD integration (GitHub Actions badge)

Tech Stack

Component	Technology
Language	Python 3.10+
Dependencies	None (stdlib only)
Terminal UI	ANSI escape codes
Data Storage	JSON (local filesystem)
Testing	unittest (stdlib)
Package Manager	pip / setuptools

Project Structure

ModelPulse-CLI/
├── src/modelpulse/
│   ├── __init__.py      # Package metadata
│   ├── cli.py           # Main CLI entry point & model database
│   ├── engine.py        # Scoring, cost analysis, trend tracking
│   ├── models.py        # Data model definitions
│   └── tui.py           # Terminal UI components
├── tests/
│   └── test_modelpulse.py  # 24 unit tests
├── data/                # User data directory (gitignored)
├── pyproject.toml       # Package configuration
├── .gitignore
├── LICENSE              # MIT License
├── CONTRIBUTING.md      # Contribution guidelines
├── logo.jpg             # Project logo
└── README.md            # This file

Contributing

We welcome contributions! Please see CONTRIBUTING.md for details.

Quick steps:

1. Fork the repository
2. Create a feature branch
3. Add/update tests
4. Submit a Pull Request

License

This project is licensed under the MIT License - see the LICENSE file for details.

---

Built with ❤️ by ModelPulse Team