metaai-sdk 3.0.0

Feature-rich Python SDK for Meta AI - Chat, Image & Video Generation powered by Llama 3

🤖 Meta AI Python SDK

Unleash the Power of Meta AI with Python 🚀

A modern, feature-rich Python SDK providing seamless access to Meta AI's cutting-edge capabilities: Chat with Llama 3, Generate Images, Create AI Videos - All Without API Keys!

🎯 Quick Start • 📖 Documentation • 💡 Examples • 🎬 Video Generation

---

✨ Why Choose This SDK?

🎯 Zero Configuration No API keys needed! Just install and start coding

⚡ Lightning Fast Optimized for performance Real-time responses

🔥 Feature Complete Chat • Images • Videos All in one SDK

🌟 Core Capabilities

Feature	Description	Status
💬 Intelligent Chat	Powered by Llama 3 with internet access	✅ Ready
📤 Image Upload	Upload & analyze images, generate similar images	✅ Ready
🎨 Image Generation	Create stunning AI-generated images	✅ Ready
🎬 Video Generation	Generate videos from text or uploaded images	✅ Ready
🔍 Image Analysis	Describe, analyze, and extract info from images	✅ Ready
🌐 Real-time Data	Get current information via Bing integration	✅ Ready
📚 Source Citations	Responses include verifiable sources	✅ Ready
🔄 Streaming Support	Real-time response streaming	✅ Ready
🔐 Auto Token Management	Automatic authentication handling	✅ Ready
🌍 Proxy Support	Route requests through proxies	✅ Ready

---

📦 Installation

SDK Only (Lightweight)

For using Meta AI as a Python library:

pip install metaai-sdk

SDK + API Server

For deploying as a REST API service:

pip install metaai-sdk[api]

From Source

git clone https://github.com/mir-ashiq/metaai-api.git
cd metaai-api
pip install -e .          # SDK only
pip install -e ".[api]"   # SDK + API server

System Requirements: Python 3.7+ • Internet Connection • That's it!

---

🚀 Quick Start

Example 1: Ask a Question

from metaai_api import MetaAI

# Initialize the AI
ai = MetaAI()

# Ask anything!
response = ai.prompt("Who won the NBA championship in 2024?")
print(response["message"])

Output:

The Boston Celtics won the 2024 NBA Championship, defeating the Dallas Mavericks
4-1 in the Finals. Jayson Tatum and Jaylen Brown led the Celtics to their 18th
championship title, the most in NBA history. The series concluded on June 17, 2024,
with the Celtics winning Game 5 at TD Garden in Boston.

Example 2: Get Stock Market Info

from metaai_api import MetaAI

ai = MetaAI()
response = ai.prompt("What is the current price of Bitcoin?")

print(f"💰 {response['message']}")
print(f"\n📚 Sources: {len(response['sources'])} references found")

Output:

💰 As of November 22, 2025, Bitcoin (BTC) is trading at approximately $97,845 USD.
The cryptocurrency has seen a 3.2% increase in the last 24 hours. Bitcoin's market
capitalization stands at around $1.93 trillion, maintaining its position as the
largest cryptocurrency by market cap.

📚 Sources: 4 references found

Example 3: Solve Math Problems

from metaai_api import MetaAI

ai = MetaAI()

# Complex calculation
question = "If I invest $10,000 at 7% annual interest compounded monthly for 5 years, how much will I have?"
response = ai.prompt(question)

print(response["message"])

Output:

With an initial investment of $10,000 at a 7% annual interest rate compounded monthly
over 5 years, you would have approximately $14,176.25.

Here's the breakdown:
- Principal: $10,000
- Interest Rate: 7% per year (0.583% per month)
- Time: 5 years (60 months)
- Compound Frequency: Monthly
- Total Interest Earned: $4,176.25
- Final Amount: $14,176.25

This calculation uses the compound interest formula: A = P(1 + r/n)^(nt)

---

💬 Chat Features

Streaming Responses

Watch responses appear in real-time, like ChatGPT:

from metaai_api import MetaAI

ai = MetaAI()

print("🤖 AI: ", end="", flush=True)
for chunk in ai.prompt("Explain quantum computing in simple terms", stream=True):
    print(chunk["message"], end="", flush=True)
print("\n")

Output:

🤖 AI: Quantum computing is like having a super-powered calculator that can solve
problems in completely new ways. Instead of regular computer bits that are either
0 or 1, quantum computers use "qubits" that can be both 0 and 1 at the same time -
imagine flipping a coin that's both heads and tails until you look at it! This
special ability allows quantum computers to process massive amounts of information
simultaneously, making them incredibly fast for specific tasks like drug discovery,
cryptography, and complex simulations.

Conversation Context

Have natural back-and-forth conversations:

from metaai_api import MetaAI

ai = MetaAI()

# First question
response1 = ai.prompt("What are the three primary colors?")
print("Q1:", response1["message"][:100])

# Follow-up question (maintains context)
response2 = ai.prompt("How do you mix them to make purple?")
print("Q2:", response2["message"][:150])

# Start fresh conversation
response3 = ai.prompt("What's the capital of France?", new_conversation=True)
print("Q3:", response3["message"][:50])

Output:

Q1: The three primary colors are Red, Blue, and Yellow. These colors cannot be created by mixing...

Q2: To make purple, you mix Red and Blue together. The exact shade of purple depends on the ratio - more red creates a reddish-purple (like magenta)...

Q3: The capital of France is Paris, located in the...

Using Proxies

Route your requests through a proxy:

from metaai_api import MetaAI

# Configure proxy
proxy = {
    'http': 'http://your-proxy-server:8080',
    'https': 'https://your-proxy-server:8080'
}

ai = MetaAI(proxy=proxy)
response = ai.prompt("Hello from behind a proxy!")
print(response["message"])

---

� REST API Server (Optional)

Deploy Meta AI as a REST API service that anyone can use! The API server auto-refreshes cookies to keep sessions alive.

Installation

pip install metaai-sdk[api]

Setup

1. Get your Meta AI cookies (see Video Generation section)
2. Create .env file:

META_AI_DATR=your_datr_cookie
META_AI_ABRA_SESS=your_abra_sess_cookie
META_AI_DPR=1
META_AI_WD=1920x1080
META_AI_REFRESH_INTERVAL_SECONDS=3600

3. Start the server:

uvicorn metaai_api.api_server:app --host 0.0.0.0 --port 8000

API Endpoints

Endpoint	Method	Description
/upload	POST	Upload images for analysis/generation
/chat	POST	Send chat messages (with/without imgs)
/image	POST	Generate images (from text or imgs)
/video	POST	Generate video (blocks until complete)
/video/async	POST	Start async video generation
/video/jobs/{job_id}	GET	Poll async job status
/healthz	GET	Health check

Example Usage

import requests

# Chat
response = requests.post("http://localhost:8000/chat", json={
    "message": "What is the capital of France?",
    "stream": False
})
print(response.json())

# Image generation
images = requests.post("http://localhost:8000/image", json={
    "prompt": "Cyberpunk cityscape at night",
    "new_conversation": False
})
print(images.json())

# Async video generation
job = requests.post("http://localhost:8000/video/async", json={
    "prompt": "Generate a video of a sunset"
})
job_id = job.json()["job_id"]

# Poll for result
status = requests.get(f"http://localhost:8000/video/jobs/{job_id}")
print(status.json())

Testing

python test_api.py

---

�🎬 Video Generation

Create AI-generated videos from text descriptions!

Setup: Get Your Cookies

1. Visit meta.ai in your browser
2. Open DevTools (F12) → Network tab
3. Refresh the page
4. Click any request → Headers → Copy Cookie value
5. Extract these values: datr, abra_sess, dpr, wd

Example 1: Generate Your First Video

from metaai_api import MetaAI

# Your browser cookies
cookies = {
    "datr": "your_datr_value_here",
    "abra_sess": "your_abra_sess_value_here",
    "dpr": "1.25",
    "wd": "1920x1080"
}

# Initialize with cookies
ai = MetaAI(cookies=cookies)

# Generate a video
result = ai.generate_video("A majestic lion walking through the African savanna at sunset")

if result["success"]:
    print("✅ Video generated successfully!")
    print(f"🎬 Generated {len(result['video_urls'])} videos (Meta AI creates 4 by default)")
    for i, url in enumerate(result['video_urls'], 1):
        print(f"   Video {i}: {url}")
    print(f"📝 Prompt: {result['prompt']}")
    print(f"🆔 Conversation ID: {result['conversation_id']}")
else:
    print("⏳ Video is still processing, try again in a moment")

Output:

[*] Fetching missing tokens (lsd, fb_dtsg) from Meta AI...
[✓] Fetched lsd: AVrvi8aHxzQ
[✓] Fetched fb_dtsg: BQAB9uRXmPEYGkC...

✅ Sending video generation request...
✅ Video generation request sent successfully!
⏳ Waiting 10 seconds before polling...
🔄 Polling for video URLs (Attempt 1/30)...
✅ Video URLs found!

✅ Video generated successfully!
🎬 Generated 4 videos (Meta AI creates 4 by default)
   Video 1: https://scontent.xx.fbcdn.net/v/t66.36240-6/video1.mp4?...
   Video 2: https://scontent.xx.fbcdn.net/v/t66.36240-6/video2.mp4?...
   Video 3: https://scontent.xx.fbcdn.net/v/t66.36240-6/video3.mp4?...
   Video 4: https://scontent.xx.fbcdn.net/v/t66.36240-6/video4.mp4?...
📝 Prompt: A majestic lion walking through the African savanna at sunset
🆔 Conversation ID: abc123-def456-ghi789

Example 2: Generate Multiple Videos

from metaai_api import MetaAI
import time

ai = MetaAI(cookies=cookies)

prompts = [
    "A futuristic city with flying cars at night",
    "Ocean waves crashing on a tropical beach",
    "Northern lights dancing over a snowy mountain"
]

videos = []
for i, prompt in enumerate(prompts, 1):
    print(f"\n🎬 Generating video {i}/{len(prompts)}: {prompt}")
    result = ai.generate_video(prompt, verbose=False)

    if result["success"]:
        videos.append(result["video_urls"][0])
        print(f"✅ Success! URL: {result['video_urls'][0][:50]}...")
    else:
        print("⏳ Still processing...")

    time.sleep(5)  # Be nice to the API

print(f"\n🎉 Generated {len(videos)} videos successfully!")

Output:

🎬 Generating video 1/3: A futuristic city with flying cars at night
✅ Success! URL: https://scontent.xx.fbcdn.net/v/t66.36240-6/1234...

🎬 Generating video 2/3: Ocean waves crashing on a tropical beach
✅ Success! URL: https://scontent.xx.fbcdn.net/v/t66.36240-6/5678...

🎬 Generating video 3/3: Northern lights dancing over a snowy mountain
✅ Success! URL: https://scontent.xx.fbcdn.net/v/t66.36240-6/9012...

🎉 Generated 3 videos successfully!

Example 3: Advanced Video Generation with Orientation

from metaai_api import MetaAI

ai = MetaAI(cookies=cookies)

# Generate video with specific orientation (default is VERTICAL)
result = ai.generate_video(
    prompt="A time-lapse of a flower blooming",
    orientation="VERTICAL",   # Options: "LANDSCAPE", "VERTICAL", "SQUARE"
    wait_before_poll=15,      # Wait 15 seconds before checking
    max_attempts=50,          # Try up to 50 times
    wait_seconds=3,           # Wait 3 seconds between attempts
    verbose=True              # Show detailed progress
)

# Generate landscape video for widescreen
result_landscape = ai.generate_video(
    prompt="Panoramic view of sunset over mountains",
    orientation="LANDSCAPE"   # Wide format (16:9)
)

if result["success"]:
    print(f"\n🎬 Your videos are ready!")
    print(f"🔗 Generated {len(result['video_urls'])} videos:")
    for i, url in enumerate(result['video_urls'], 1):
        print(f"   Video {i}: {url}")
    print(f"⏱️ Generated at: {result['timestamp']}")

Supported Video Orientations:

• "LANDSCAPE" - Wide/horizontal (16:9) - ideal for widescreen, cinematic content
• "VERTICAL" - Tall/vertical (9:16) - ideal for mobile, stories, reels (default)
• "SQUARE" - Equal dimensions (1:1) - ideal for social posts

📖 **Full Video Guide:** See [VIDEO_GENERATION_README.md](https://github.com/mir-ashiq/metaai-api/blob/main/VIDEO_GENERATION_README.md) for complete documentation!

---

## 📤 Image Upload & Analysis

Upload images to Meta AI for analysis, similar image generation, and video creation:

### Upload & Analyze Images

```python
from metaai_api import MetaAI

# Initialize with Facebook cookies (required for image operations)
ai = MetaAI(cookies={
    "datr": "your_datr_cookie",
    "abra_sess": "your_abra_sess_cookie"
})

# Step 1: Upload an image
result = ai.upload_image("path/to/image.jpg")

if result["success"]:
    media_id = result["media_id"]
    metadata = {
        'file_size': result['file_size'],
        'mime_type': result['mime_type']
    }

    # Step 2: Analyze the image
    response = ai.prompt(
        message="What do you see in this image? Describe it in detail.",
        media_ids=[media_id],
        attachment_metadata=metadata
    )
    print(f"🔍 Analysis: {response['message']}")

    # Step 3: Generate similar images
    response = ai.prompt(
        message="Create a similar image in watercolor painting style",
        media_ids=[media_id],
        attachment_metadata=metadata,
        is_image_generation=True
    )
    print(f"🎨 Generated {len(response['media'])} similar images")

    # Step 4: Generate video from image
    video = ai.generate_video(
        prompt="generate a video with zoom in effect on this image",
        media_ids=[media_id],
        attachment_metadata=metadata
    )
    if video["success"]:
        print(f"🎬 Video: {video['video_urls'][0]}")

Output:

🔍 Analysis: The image captures a serene lake scene set against a majestic mountain backdrop. In the foreground, there's a small, golden-yellow wooden boat with a bright yellow canopy floating on calm, glass‑like water...

🎨 Generated 4 similar images

🎬 Video: https://scontent.fsxr1-2.fna.fbcdn.net/o1/v/t6/f2/m421/video.mp4

📖 Full Image Upload Guide: See IMAGE_UPLOAD_README.md for complete documentation!

---

🎨 Image Generation

Generate AI-powered images with customizable orientations (requires Facebook authentication):

from metaai_api import MetaAI

# Initialize with Facebook credentials
ai = MetaAI(fb_email="[e-mail]", fb_password="your_password")

# Generate images with default orientation (VERTICAL)
response = ai.prompt("Generate an image of a cyberpunk cityscape at night with neon lights")

# Or specify orientation explicitly
response_landscape = ai.prompt(
    "Generate an image of a panoramic mountain landscape",
    orientation="LANDSCAPE"  # Options: "LANDSCAPE", "VERTICAL", "SQUARE"
)

response_vertical = ai.prompt(
    "Generate an image of a tall waterfall",
    orientation="VERTICAL"  # Tall/portrait format (default)
)

response_square = ai.prompt(
    "Generate an image of a centered mandala pattern",
    orientation="SQUARE"  # Square format (1:1)
)

# Display results (Meta AI generates 4 images by default)
print(f"🎨 Generated {len(response['media'])} images:")
for i, image in enumerate(response['media'], 1):
    print(f"  Image {i}: {image['url']}")
    print(f"  Prompt: {image['prompt']}")

Supported Orientations:

• "LANDSCAPE" - Wide/horizontal format (16:9) - ideal for panoramas, landscapes
• "VERTICAL" - Tall/vertical format (9:16) - ideal for portraits, mobile content (default)
• "SQUARE" - Equal dimensions (1:1) - ideal for social media, profile images

Output:

🎨 Generated 4 images:
  Image 1: https://scontent.xx.fbcdn.net/o1/v/t0/f1/m247/img1.jpeg
  Prompt: a cyberpunk cityscape at night with neon lights

  Image 2: https://scontent.xx.fbcdn.net/o1/v/t0/f1/m247/img2.jpeg
  Prompt: a cyberpunk cityscape at night with neon lights

  Image 3: https://scontent.xx.fbcdn.net/o1/v/t0/f1/m247/img3.jpeg
  Prompt: a cyberpunk cityscape at night with neon lights

  Image 4: https://scontent.xx.fbcdn.net/o1/v/t0/f1/m247/img4.jpeg
  Prompt: a cyberpunk cityscape at night with neon lights

---

💡 Examples

Explore working examples in the examples/ directory:

File	Description	Features
📄 image_workflow_complete.py	Complete image workflow	Upload, analyze, generate images/video
📄 simple_example.py	Quick start guide	Basic chat + video generation
📄 video_generation.py	Video generation	Multiple examples, error handling
📄 test_example.py	Testing suite	Validation and testing

Run an Example

# Clone the repository
git clone https://github.com/mir-ashiq/metaai-api.git
cd meta-ai-python

# Run simple example
python examples/simple_example.py

# Run video generation examples
python examples/video_generation.py

---

📖 Documentation

📚 Complete Guides

Document	Description
📘 Image Upload Guide	Complete image upload documentation
📘 Video Generation Guide	Complete video generation documentation
📙 Quick Reference	Fast lookup for common tasks
📙 Quick Usage	Image upload quick reference
📗 Architecture Guide	Technical architecture details
📕 Contributing Guide	How to contribute to the project
📔 Changelog	Version history and updates
📓 Security Policy	Security best practices

🔧 API Reference

MetaAI Class

class MetaAI:
    def __init__(
        self,
        fb_email: Optional[str] = None,
        fb_password: Optional[str] = None,
        cookies: Optional[dict] = None,
        proxy: Optional[dict] = None
    )

Methods:

• prompt(message, stream=False, new_conversation=False)

  • Send a chat message
  • Returns: dict with message, sources, and media

• generate_video(prompt, wait_before_poll=10, max_attempts=30, wait_seconds=5, verbose=True)

  • Generate a video from text
  • Returns: dict with success, video_urls, conversation_id, prompt, timestamp

VideoGenerator Class

from metaai_api import VideoGenerator

# Direct video generation
generator = VideoGenerator(cookies_str="your_cookies_as_string")
result = generator.generate_video("your prompt here")

# One-liner generation
result = VideoGenerator.quick_generate(
    cookies_str="your_cookies",
    prompt="your prompt"
)

---

🎯 Use Cases

1. Research Assistant

ai = MetaAI()
research = ai.prompt("Summarize recent breakthroughs in fusion energy")
print(research["message"])
# Get cited sources
for source in research["sources"]:
    print(f"📌 {source['title']}: {source['link']}")

2. Content Creation

ai = MetaAI(cookies=cookies)

# Generate video content
promo_video = ai.generate_video("Product showcase with smooth camera movements")

# Generate images
thumbnails = ai.prompt("Generate a YouTube thumbnail for a tech review video")

3. Educational Tool

ai = MetaAI()

# Explain complex topics
explanation = ai.prompt("Explain blockchain technology to a 10-year-old")

# Get homework help
solution = ai.prompt("Solve: 2x + 5 = 13, show steps")

4. Real-time Information

ai = MetaAI()

# Current events
news = ai.prompt("What are the top technology news today?")

# Sports scores
scores = ai.prompt("Latest Premier League scores")

# Market data
stocks = ai.prompt("Current S&P 500 index value")

---

🛠️ Advanced Configuration

Environment Variables

Store credentials securely:

# .env file
META_AI_DATR=your_datr_value
META_AI_ABRA_SESS=your_abra_sess_value
META_AI_DPR=1.25
META_AI_WD=1920x1080

Load in Python:

import os
from dotenv import load_dotenv
from metaai_api import MetaAI

load_dotenv()

cookies = {
    "datr": os.getenv("META_AI_DATR"),
    "abra_sess": os.getenv("META_AI_ABRA_SESS"),
    "dpr": os.getenv("META_AI_DPR"),
    "wd": os.getenv("META_AI_WD")
}

ai = MetaAI(cookies=cookies)

Error Handling

from metaai_api import MetaAI

ai = MetaAI(cookies=cookies)

try:
    result = ai.generate_video("Your prompt")

    if result["success"]:
        print(f"✅ Video: {result['video_urls'][0]}")
    else:
        print("⏳ Video still processing, try again later")

except ValueError as e:
    print(f"❌ Configuration error: {e}")
except ConnectionError as e:
    print(f"❌ Network error: {e}")
except Exception as e:
    print(f"❌ Unexpected error: {e}")

---

🌟 Project Structure

meta-ai-python/
│
├── 📁 src/metaai_api/        # Core package
│   ├── __init__.py            # Package initialization
│   ├── main.py                # MetaAI class
│   ├── video_generation.py    # Video generation
│   ├── client.py              # Client utilities
│   ├── utils.py               # Helper functions
│   └── exceptions.py          # Custom exceptions
│
├── 📁 examples/               # Usage examples
│   ├── simple_example.py      # Quick start
│   ├── video_generation.py    # Video examples
│   └── test_example.py        # Testing
│
├── 📁 .github/                # GitHub configuration
│   ├── workflows/             # CI/CD pipelines
│   └── README.md
│
├── 📄 README.md               # This file
├── 📄 VIDEO_GENERATION_README.md
├── 📄 QUICK_REFERENCE.md
├── 📄 ARCHITECTURE.md
├── 📄 CONTRIBUTING.md
├── 📄 CHANGELOG.md
├── 📄 SECURITY.md
├── 📄 LICENSE                 # MIT License
├── 📄 setup.py                # Package setup
├── 📄 pyproject.toml          # Project metadata
└── 📄 requirements.txt        # Dependencies

---

🤝 Contributing

We welcome contributions! Here's how you can help:

1. 🐛 Report Bugs - Open an issue
2. 💡 Suggest Features - Share your ideas
3. 📝 Improve Docs - Help us document better
4. 🔧 Submit PRs - Fix bugs or add features

See CONTRIBUTING.md for detailed guidelines.

---

📜 License

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

⚖️ Disclaimer

This project is an independent implementation and is not officially affiliated with Meta Platforms, Inc. or any of its affiliates.

• ✅ Educational and development purposes
• ✅ Use responsibly and ethically
• ✅ Comply with Meta's Terms of Service
• ✅ Respect usage limits and policies

Llama 3 License: Visit llama.com/llama3/license for Llama 3 usage terms.

---

🙏 Acknowledgments

• Meta AI - For providing the AI capabilities
• Llama 3 - The powerful language model
• Open Source Community - For inspiration and support

---

📞 Support & Community

• 💬 Questions? GitHub Discussions
• 🐛 Bug Reports GitHub Issues
• 📧 Contact [e-mail]
• ⭐ Star us on GitHub

---

🚀 Quick Links

Resource	Link
📦 PyPI Package	pypi.org/project/metaai_api
🐙 GitHub Repository	github.com/mir-ashiq/meta-ai-python
📖 Full Documentation	Video Guide • Quick Ref
💬 Get Help	Issues • Discussions

---

_{Meta AI Python SDK v2.0.0 | Made with ❤️ by mir-ashiq | MIT License}

⭐ Star this repo if you find it useful!