Google Gemini

Google Gemini is a powerful multimodal AI model that excels at understanding and generating text, images, and code. This guide covers how to use Gemini through EasilyAI.

Getting Started

API Key Setup

1. Go to Google AI Studio
2. Create an API key
3. Set it as an environment variable or use it directly

export GEMINI_API_KEY="your-gemini-api-key"

Basic Usage

from easilyai import create_app

# Create Gemini app
app = create_app("Gemini", "gemini", "your-gemini-api-key", "gemini-1.5-flash")

# Generate text
response = app.request("Explain quantum computing in simple terms")
print(response)

Available Models

Gemini 1.5 Flash

• Model ID: gemini-1.5-flash
• Best for: Fast responses, general tasks
• Context window: 1M tokens
• Strengths: Speed, efficiency, cost-effective

flash_app = create_app("GeminiFlash", "gemini", "your-key", "gemini-1.5-flash")
response = flash_app.request("Quick question: What is 2+2?")

Gemini 1.5 Pro

• Model ID: gemini-1.5-pro
• Best for: Complex reasoning, analysis
• Context window: 2M tokens
• Strengths: Advanced reasoning, multimodal capabilities

pro_app = create_app("GeminiPro", "gemini", "your-key", "gemini-1.5-pro")
analysis = pro_app.request("Analyze the economic impact of renewable energy adoption")

Gemini 1.0 Pro

• Model ID: gemini-1.0-pro
• Best for: General purpose tasks
• Context window: 30K tokens
• Strengths: Balanced performance

pro_app = create_app("GeminiPro", "gemini", "your-key", "gemini-1.0-pro")
response = pro_app.request("Write a short story about artificial intelligence")

Parameters

Text Generation Parameters

response = app.request(
    "Write a creative story",
    temperature=0.7,           # Controls randomness (0.0 to 1.0)
    max_output_tokens=1000,    # Maximum tokens to generate
    top_p=0.8,                 # Nucleus sampling
    top_k=40                   # Top-k sampling
)

Safety Settings

Gemini includes built-in safety filters. You can adjust these through the API:

# Note: Safety settings are handled internally by Google's API
response = app.request(
    "Write a safety-conscious story about AI",
    safety_settings={
        "HARM_CATEGORY_HARASSMENT": "BLOCK_MEDIUM_AND_ABOVE",
        "HARM_CATEGORY_HATE_SPEECH": "BLOCK_MEDIUM_AND_ABOVE"
    }
)

Use Cases

Creative Writing

Gemini excels at creative tasks:

from easilyai import create_app

creative_app = create_app("Creative", "gemini", "your-key", "gemini-1.5-pro")

# Story generation
story = creative_app.request(
    "Write a compelling short story about a time traveler who discovers "
    "they can only travel to moments of great historical significance. "
    "Make it 500 words and include dialogue."
)

# Poetry
poem = creative_app.request(
    "Write a haiku about the relationship between technology and nature"
)

# Screenwriting
screenplay = creative_app.request(
    "Write a short dialogue between two AI systems discussing consciousness"
)

Code Generation

Gemini is excellent for programming tasks:

from easilyai import create_app

coder_app = create_app("Coder", "gemini", "your-key", "gemini-1.5-pro")

# Function generation
function_code = coder_app.request(
    "Write a Python function that implements a binary search algorithm. "
    "Include proper error handling, type hints, and docstrings."
)

# Code explanation
explanation = coder_app.request(
    "Explain this Python code step by step:\n"
    "def fibonacci(n):\n"
    "    if n <= 1:\n"
    "        return n\n"
    "    return fibonacci(n-1) + fibonacci(n-2)"
)

# Code review
review = coder_app.request(
    "Review this code for potential improvements:\n"
    "def calculate_average(numbers):\n"
    "    return sum(numbers) / len(numbers)"
)

Data Analysis

Use Gemini for analyzing data and generating insights:

from easilyai import create_app

analyst_app = create_app("Analyst", "gemini", "your-key", "gemini-1.5-pro")

# Data analysis
analysis = analyst_app.request(
    "Analyze this sales data and provide insights:\n"
    "January: $45,000\n"
    "February: $52,000\n"
    "March: $48,000\n"
    "April: $61,000\n"
    "May: $59,000\n"
    "June: $67,000\n"
    "What trends do you see? What recommendations would you make?"
)

# Research synthesis
research = analyst_app.request(
    "Synthesize the key findings from recent research on climate change impacts "
    "on agriculture. Focus on the most significant threats and adaptation strategies."
)

Educational Content

Create educational materials:

from easilyai import create_app

teacher_app = create_app("Teacher", "gemini", "your-key", "gemini-1.5-flash")

# Lesson plans
lesson = teacher_app.request(
    "Create a lesson plan for teaching 8th graders about photosynthesis. "
    "Include learning objectives, activities, and assessment methods."
)

# Explanations
explanation = teacher_app.request(
    "Explain the concept of machine learning to a 10-year-old using simple "
    "analogies and examples they can relate to."
)

# Quiz generation
quiz = teacher_app.request(
    "Create a 10-question multiple choice quiz about the solar system. "
    "Include answer explanations for each question."
)

Advanced Features

Long Context Processing

Gemini 1.5 models support very long contexts:

from easilyai import create_app

long_context_app = create_app("LongContext", "gemini", "your-key", "gemini-1.5-pro")

# Process large documents
large_document = """
[Insert very long document here - up to 1M tokens for Flash, 2M for Pro]
"""

summary = long_context_app.request(
    f"Please summarize this document and extract the key points:\n\n{large_document}"
)

Multimodal Capabilities

While EasilyAI focuses on text, Gemini supports multimodal inputs through Google's API:

# Note: Direct multimodal support would require additional implementation
# This is a text-based approach to describe images
image_analysis = app.request(
    "Describe what would be in an image that shows 'a bustling city street at night "
    "with neon signs and reflections on wet pavement'"
)

Chain of Thought Reasoning

Leverage Gemini's reasoning capabilities:

from easilyai import create_app

reasoning_app = create_app("Reasoning", "gemini", "your-key", "gemini-1.5-pro")

# Complex problem solving
solution = reasoning_app.request(
    "Solve this step by step:\n"
    "A company has 100 employees. 60% work in engineering, 25% in sales, "
    "and the rest in administration. If the company grows by 50% and maintains "
    "the same proportions, how many new employees will be needed in each department?"
)

# Logical reasoning
logic = reasoning_app.request(
    "Think through this logical puzzle step by step:\n"
    "All cats are animals. Some animals are pets. Therefore, some cats are pets. "
    "Is this reasoning valid? Explain your analysis."
)

Optimization Tips

Model Selection

Choose the right model for your use case:

# For quick responses and high throughput
flash_app = create_app("Quick", "gemini", "your-key", "gemini-1.5-flash")

# For complex reasoning and analysis
pro_app = create_app("Complex", "gemini", "your-key", "gemini-1.5-pro")

# For budget-conscious applications
budget_app = create_app("Budget", "gemini", "your-key", "gemini-1.0-pro")

Performance Optimization

from easilyai import create_app

app = create_app("Optimized", "gemini", "your-key", "gemini-1.5-flash")

# Use lower temperature for consistent results
consistent_response = app.request(
    "Summarize this article",
    temperature=0.1
)

# Use higher temperature for creative tasks
creative_response = app.request(
    "Write a creative story",
    temperature=0.8
)

# Limit output tokens for concise responses
brief_response = app.request(
    "Explain quantum computing briefly",
    max_output_tokens=100
)

Prompt Engineering

Optimize your prompts for better results:

from easilyai import create_app

app = create_app("Optimized", "gemini", "your-key", "gemini-1.5-pro")

# Structured prompts work well
structured_prompt = """
Task: Write a product description
Product: Wireless headphones
Requirements:
- Highlight key features
- Target audience: professionals
- Tone: professional yet engaging
- Length: 100-150 words

Please format the response with clear sections for features and benefits.
"""

description = app.request(structured_prompt)

Error Handling

Common Errors

from easilyai import create_app
from easilyai.exceptions import EasilyAIException

app = create_app("Gemini", "gemini", "your-key", "gemini-1.5-flash")

try:
    response = app.request("Your prompt here")
    print(response)
except EasilyAIException as e:
    error_msg = str(e).lower()
    
    if "api key" in error_msg:
        print("Invalid API key. Please check your Gemini API key.")
    elif "quota" in error_msg:
        print("API quota exceeded. Please check your usage limits.")
    elif "safety" in error_msg:
        print("Content was blocked by safety filters. Try rephrasing your prompt.")
    else:
        print(f"Gemini API error: {e}")

Rate Limiting

import time
from easilyai import create_app

def gemini_request_with_retry(prompt, max_retries=3):
    app = create_app("Gemini", "gemini", "your-key", "gemini-1.5-flash")
    
    for attempt in range(max_retries):
        try:
            return app.request(prompt)
        except EasilyAIException as e:
            if "rate limit" in str(e).lower() and attempt < max_retries - 1:
                wait_time = 2 ** attempt  # Exponential backoff
                time.sleep(wait_time)
                continue
            else:
                raise e
    
    return None

Best Practices

1. Choose the Right Model

• Gemini 1.5 Flash: Fast responses, high throughput
• Gemini 1.5 Pro: Complex reasoning, long context
• Gemini 1.0 Pro: Balanced performance

2. Optimize Prompts

# Good: Specific and clear
good_prompt = "Write a 200-word product description for wireless earbuds targeting fitness enthusiasts"

# Better: Structured with examples
better_prompt = """
Write a product description for wireless earbuds.
Target audience: Fitness enthusiasts
Key features: Waterproof, 12-hour battery, noise cancellation
Tone: Energetic and motivational
Length: 200 words
Format: Include headline, features list, and call-to-action
"""

3. Use Safety Features

Gemini includes built-in safety filters that help ensure responsible AI use.

4. Monitor Usage

Track your API usage to manage costs:

import time
from collections import defaultdict

class GeminiUsageTracker:
    def __init__(self):
        self.usage_log = defaultdict(int)
    
    def track_request(self, model, tokens_used):
        self.usage_log[model] += tokens_used
    
    def get_usage_summary(self):
        return dict(self.usage_log)

Comparison with Other Services

Gemini vs GPT-4

• Gemini: Longer context window, integrated safety features
• GPT-4: More established ecosystem, wider adoption

Gemini vs Claude

• Gemini: Better for code generation, multimodal capabilities
• Claude: Better for reasoning, analysis tasks

When to Use Gemini

• Long document processing
• Code generation and analysis
• Creative writing projects
• Educational content creation
• Research and analysis tasks

Google Gemini offers powerful capabilities through EasilyAI's simple interface. Its long context windows and multimodal capabilities make it excellent for complex, context-heavy tasks.