We're excited to announce that Gemini 3 Series API, Google's most advanced AI model family, is now available on PoYo. Access Gemini 3 Flash Preview and Gemini 3 Pro Preview through a unified API with no waitlist, instant access, and competitive pricing.
What Makes Gemini 3 Series Revolutionary
Gemini 3 represents Google's breakthrough in AI capabilities, featuring the largest context window among frontier models, dynamic thinking levels, and state-of-the-art benchmark performance.
Gemini 3 Flash Preview - Speed Champion
Model ID: gemini-3-flash-preview
Gemini 3 Flash Preview is Google's fastest frontier model, designed for rapid iteration and high-volume tasks:
- 3x Faster Than Gemini 2.5 Pro — Dramatically reduced latency for real-time applications
- 78% SWE-bench Verified — Strong coding performance for agentic development
- 1M Token Context Window — Process entire codebases in a single request
- Dynamic Thinking Levels — Configure reasoning depth per-request (minimal/low/medium/high)
- Multimodal Input — Text, images, audio, video, and PDF processing
Pricing: $0.4/1M input tokens, $2.4/1M output tokens
Gemini 3 Pro Preview - Reasoning Leader
Model ID: gemini-3-pro-preview
Gemini 3 Pro Preview leads the industry in reasoning capabilities:
- LMArena #1 with 1501 Elo — The highest-scoring model on the crowdsourced leaderboard
- 91.9% GPQA Diamond — PhD-level reasoning accuracy
- 37.5% Humanity's Last Exam — Exceptional performance on frontier benchmarks
- MoE Architecture (1T+ Parameters) — Only 15-20B parameters activated per query for efficiency
- 1M Token Context Window — Same massive context as Flash
Pricing: $0.8/1M input tokens, $4.8/1M output tokens
Key Features Across Gemini 3 Models
1M Token Context Window
Gemini 3 offers the largest context window among frontier models — 1 million tokens. This enables:
- Full Repository Analysis — Process entire codebases in a single request
- Long Document Processing — Analyze complete research papers, legal documents, and reports
- Multi-File Context — Maintain context across hundreds of files simultaneously
- Automatic Context Caching — Efficient handling of repeated content
Dynamic Thinking Levels
Unlike fixed reasoning modes, Gemini 3 offers configurable thinking depth through the thinking_level parameter:
| Level | Use Case | Latency | Cost |
|---|---|---|---|
| Minimal | Simple queries, classification | Lowest | Lowest |
| Low | Standard tasks, quick responses | Low | Low |
| Medium | Moderate complexity, balanced | Medium | Medium |
| High | Complex reasoning, deep analysis | Higher | Higher |
This flexibility lets you optimize cost and latency per-request based on task complexity.
Multimodal Understanding
Gemini 3 supports the richest multimodal input among frontier models:
- Text — Natural language prompts and instructions
- Images — Visual analysis with configurable resolution (low/medium/high)
- Audio — Speech and audio file processing
- Video — Video understanding and analysis
- PDF — Native PDF document processing
Use the media_resolution parameter to control vision processing and optimize token usage.
Built-in Tools
Gemini 3 natively supports powerful built-in tools without external integrations:
- Google Search — Real-time web information retrieval
- Code Execution — Run and test code in-context
- File Search — Document retrieval and analysis
- URL Context — Web page analysis and extraction
These tools enable sophisticated agentic workflows directly within the API.
Dual API Endpoint Support
PoYo supports two endpoints for Gemini 3 models:
/v1/chat/completions— OpenAI-compatible endpoint for SDK compatibility/v1beta/models/{model}:generateContent— Gemini Native Format for full feature access
Both endpoints use the same API key. Choose based on your integration requirements.
Pricing Comparison
| Model | Input Price | Output Price | Best For |
|---|---|---|---|
| Gemini 3 Flash Preview | $0.4/1M | $2.4/1M | Fast coding & agents |
| Gemini 3 Pro Preview | $0.8/1M | $4.8/1M | Deep reasoning |
All models include:
- 1M token context window
- Dynamic thinking levels
- Multimodal input (text, image, audio, video, PDF)
- Built-in tools (Google Search, Code Execution, File Search)
- /v1/chat/completions and Gemini Native Format endpoints
No subscription fees — pay only for what you use.
Getting Started with Gemini 3 API
Step 1: Get API Access
Create your PoYo account and generate your API key in under 2 minutes. No waitlist, no approval process required.
Step 2: Try the Playground
Experience Gemini 3's dynamic thinking and multimodal capabilities through our interactive web interface. Test different thinking levels and configurations.
Step 3: Integrate via API
OpenAI-Compatible Endpoint:
curl --request POST \
--url https://api.poyo.ai/v1/chat/completions \
--header 'Authorization: Bearer YOUR_API_KEY' \
--header 'Content-Type: application/json' \
--data '{
"model": "gemini-3-flash-preview",
"messages": [
{"role": "user", "content": "Explain the theory of relativity."}
]
}'
Gemini Native Format:
curl --request POST \
--url https://api.poyo.ai/v1beta/models/gemini-3-flash-preview:generateContent \
--header 'Authorization: Bearer YOUR_API_KEY' \
--header 'Content-Type: application/json' \
--data '{
"contents": [
{"role": "user", "parts": [{"text": "Explain the theory of relativity."}]}
],
"generationConfig": {
"temperature": 1.0,
"maxOutputTokens": 4096
}
}'
Use Cases
Large Codebase Analysis
Leverage the 1M token context window to analyze entire repositories in a single request. Gemini 3 Flash achieves 78% on SWE-bench Verified, making it ideal for code review, migration planning, and refactoring.
Research & Knowledge Work
Gemini 3 Pro's LMArena-leading reasoning combined with built-in Google Search enables powerful research workflows. Analyze complex datasets, academic papers, and multi-source information with PhD-level accuracy.
AI Agent Development
Build sophisticated agents with Gemini 3's built-in tools — Google Search for real-time data, Code Execution for testing, and Function Calling for integrations. Dynamic thinking levels let you optimize cost and latency per-request.
Enterprise Document Processing
Process massive document collections with the 1M token context. Analyze regulatory filings, market reports, and internal data with multimodal support for text, charts, PDFs, and visual content.
Multimodal Content Creation
Create rich content workflows with Gemini 3's support for text, images, audio, and video input. Generate descriptions, analyze media, and build multimodal applications.
Why Choose PoYo for Gemini 3 API
Instant Access
Start using Gemini 3 API in under 2 minutes. No waitlist, no approval process, no enterprise contracts required.
Most Affordable Pricing
Access Gemini 3 Flash starting from just $0.4/1M input tokens — the most cost-effective frontier model on the market.
Production-Grade Reliability
99.9% uptime SLA with automatic failover and traffic spike handling. Built for enterprise applications.
OpenAI SDK Compatible
Use your existing OpenAI SDK code by simply changing the base URL. Or use Gemini Native Format for full feature access.
Ready to start building with Gemini 3? Get your API key and access Google's most capable models today.
For developers ready to integrate, explore our comprehensive API documentation with examples in Python, JavaScript, and cURL.
Stay at the forefront of AI innovation by following our blog and exploring more powerful AI models on PoYo.