Gemini Series
The Google Gemini series is known for multimodality and cost-effectiveness, and also offers web-connected tiers with built-in Google Search.
Available Models
| Model | Positioning |
|---|---|
gemini-3.5-flash(-high / -low) | Mainstream tier, fast and cost-effective |
gemini-3.1-pro(-high / -low / -preview) | Premium tier, for complex tasks and long context |
gemini-3-flash | Previous-generation fast tier |
gemini-2.5-flash-lite | Lightest-weight tier |
*-google-search variants | Built-in web search, see Web Search |
See the real-time pricing on the Model Marketplace.
How to Use
- OpenAI compatible: just replace
model, no other changes needed; - Google native generateContent: multimodal inputs (images/video/audio) and native features, see Gemini Native Calls.
Model Selection Recommendations
- High-volume text processing:
gemini-3.5-flash, with clear advantages in speed and per-call cost; - Long documents / complex analysis:
gemini-3.1-pro(use-highfor the highest setting); - Time-sensitive Q&A:
gemini-3.5-flash-google-search; - Image/video analysis: reuse the same series directly, see Image Understanding and Video Understanding;
- Gemini image generation models (Nano Banana class) see Image and Video Generation Models.