Google's Gemini models offer exceptional multimodal capabilities, advanced reasoning, and competitive pricing. This guide covers how to integrate Gemini 2.5 Pro, Gemini 2.5 Flash, and the latest Gemini 3 Pro through Crazyrouter's API.
Supported Gemini Models#
| Model | Input ($/1M) | Output ($/1M) | Best For |
|---|---|---|---|
| gemini-3-pro-preview | $1.25 | $7.50 | Latest generation |
| gemini-3-pro-preview-thinking | $1.25 | $7.50 | Extended reasoning |
| gemini-3-pro-image-preview | $1.25 | $7.50 | Image generation |
| gemini-3-flash-preview | $0.15 | $0.60 | Fast responses |
| gemini-2.5-pro | $1.25 | $10.00 | Long documents |
| gemini-2.5-pro-thinking | $1.25 | $10.00 | Complex reasoning |
| gemini-2.5-flash | $0.30 | $2.50 | Production workloads |
| gemini-2.5-flash-lite | $0.10 | $0.40 | Cost-effective |
| gemini-2.5-flash-image | $0.30 | $2.50 | Image generation |
Quick Start with OpenAI-Compatible API#
The easiest way to use Gemini models is through the OpenAI-compatible endpoint:
python
from openai import OpenAI
client = OpenAI(
api_key="sk-your-api-key",
base_url="https://crazyrouter.com/v1",
default_headers={
"User-Agent": "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36"
}
)
response = client.chat.completions.create(
model="gemini-2.5-pro",
messages=[
{"role": "user", "content": "Explain the theory of relativity in simple terms."}
]
)
print(response.choices[0].message.content)
Native Gemini API Format#
For advanced features, use the native Gemini API format:
Text Generation#
bash
curl -X POST "https://crazyrouter.com/v1beta/models/gemini-2.5-pro:generateContent" \
-H "Content-Type: application/json" \
-H "Authorization: Bearer sk-your-api-key" \
-H "User-Agent: Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36" \
-d '{
"contents": [{
"parts": [{"text": "Write a poem about artificial intelligence."}]
}]
}'
Streaming Text Generation#
bash
curl -X POST "https://crazyrouter.com/v1beta/models/gemini-3-pro-preview:streamGenerateContent" \
-H "Content-Type: application/json" \
-H "Authorization: Bearer sk-your-api-key" \
-H "User-Agent: Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36" \
-d '{
"contents": [{
"parts": [{"text": "Explain quantum computing step by step."}]
}]
}'
Extended Thinking with Gemini#
Gemini models support extended thinking for complex reasoning:
python
# Gemini 2.5 Pro with thinking
response = client.chat.completions.create(
model="gemini-2.5-pro-thinking",
messages=[
{"role": "user", "content": "Solve this complex math problem step by step: Find all prime numbers p such that p^2 + 2 is also prime."}
]
)
# Gemini 3 Pro with thinking
response = client.chat.completions.create(
model="gemini-3-pro-preview-thinking",
messages=[
{"role": "user", "content": "Analyze this algorithm's time complexity."}
]
)
Image Generation with Gemini#
Gemini 2.5 Flash Image#
bash
curl -X POST "https://crazyrouter.com/v1beta/models/gemini-2.5-flash-image:generateContent" \
-H "Content-Type: application/json" \
-H "Authorization: Bearer sk-your-api-key" \
-H "User-Agent: Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36" \
-d '{
"contents": [{
"parts": [{"text": "Generate an image of a futuristic city at sunset"}]
}],
"generationConfig": {
"responseModalities": ["image", "text"]
}
}'
Gemini 3 Pro Image (with aspect ratio control)#
bash
curl -X POST "https://crazyrouter.com/v1beta/models/gemini-3-pro-image-preview:generateContent" \
-H "Content-Type: application/json" \
-H "Authorization: Bearer sk-your-api-key" \
-H "User-Agent: Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36" \
-d '{
"contents": [{
"parts": [{"text": "A serene Japanese garden with cherry blossoms"}]
}],
"generationConfig": {
"responseModalities": ["image"],
"aspectRatio": "16:9"
}
}'
Google Search Integration#
Gemini models can access real-time information via Google Search:
bash
curl -X POST "https://crazyrouter.com/v1beta/models/gemini-2.5-flash:generateContent" \
-H "Content-Type: application/json" \
-H "Authorization: Bearer sk-your-api-key" \
-H "User-Agent: Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36" \
-d '{
"contents": [{
"parts": [{"text": "What are the latest developments in AI as of today?"}]
}],
"tools": [{
"googleSearch": {}
}]
}'
Flash Lite Models (Cost-Effective)#
For high-volume, cost-sensitive applications:
python
# Standard Flash Lite
response = client.chat.completions.create(
model="gemini-2.5-flash-lite",
messages=[{"role": "user", "content": "Summarize this text briefly."}]
)
# Flash Lite with thinking
response = client.chat.completions.create(
model="gemini-2.5-flash-lite-thinking",
messages=[{"role": "user", "content": "Solve this logic puzzle."}]
)
Python SDK Example#
python
from openai import OpenAI
client = OpenAI(
api_key="sk-your-api-key",
base_url="https://crazyrouter.com/v1",
default_headers={
"User-Agent": "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36"
}
)
# Gemini 3 Pro for complex reasoning
response = client.chat.completions.create(
model="gemini-3-pro-preview",
messages=[
{"role": "system", "content": "You are a helpful AI assistant with expertise in science."},
{"role": "user", "content": "Explain CRISPR gene editing and its potential applications."}
],
temperature=0.7,
max_tokens=2000
)
print(response.choices[0].message.content)
Pricing Comparison#
| Model | Input ($/1M) | Output ($/1M) |
|---|---|---|
| gemini-3-pro-preview | $1.25 | $7.50 |
| gemini-2.5-pro | $1.25 | $10.00 |
| gemini-2.5-flash | $0.30 | $2.50 |
| gemini-2.5-flash-lite | $0.10 | $0.40 |
| gemini-flash-latest | $0.30 | $2.50 |
Pricing Disclaimer: The prices shown in this article are for demonstration purposes only and may change at any time. Actual billing will be based on the real-time prices displayed when you make your request.
Best Practices#
- Use Flash for production: Gemini 2.5 Flash offers excellent performance at lower cost
- Enable thinking for complex tasks: Use thinking-enabled models for math and reasoning
- Leverage multimodal capabilities: Gemini excels at image and video understanding
- Use streaming for long responses: Better user experience with real-time output
Model Selection Guide#
| Use Case | Recommended Model |
|---|---|
| General chat | gemini-2.5-flash |
| Complex reasoning | gemini-2.5-pro-thinking or gemini-3-pro-preview-thinking |
| Image generation | gemini-2.5-flash-image or gemini-3-pro-image-preview |
| Cost-sensitive | gemini-2.5-flash-lite |
| Latest features | gemini-3-pro-preview |
Getting Started#
- Sign up at Crazyrouter
- Create an API key in the console
- Choose between OpenAI-compatible or native Gemini format
- Start building with Gemini models
For questions, contact support@crazyrouter.com
Implementation Guides
Gemini Native FormatUse Gemini native generateContent requests through Crazyrouter.Quick Start GuideMake the first Crazyrouter API call and validate your setup.API EndpointsChoose the correct base URL for OpenAI-compatible, Claude, and Gemini clients.Reasoning ModelsChoose the right protocol and fields for thinking and reasoning workloads.
