Available Google AI Models Overview.
Gemini
The following Gemini models are available in the Giselle workspace. Each model offers specific capabilities and features to match your use case requirements.
| Models | Generate Text | Web Search | Reasoning | Input PDF | Input Image | Input Audio | Input Video | Context Window | Plan@Giselle |
|---|---|---|---|---|---|---|---|---|---|
| gemini-2.5-pro-exp-03-25 | ✅ | ✅ | ✅ | ✅ | ✅ | ✅ | ✅ | 1M tokens (2M soon) | Free |
| gemini-2.0-flash-001 | ✅ | ✅ | ❌ | ✅ | ✅ | ✅ | ✅ | 1M tokens | Pro |
| gemini-2.0-flash-thinking-exp-01-21 | ✅ | ❌ | ✅ | ✅ | ✅ | ✅ | ✅ | 1M tokens | Free |
| gemini-2.0-pro-exp-02-05 | ✅ | ✅ | ❌ | ✅ | ✅ | ✅ | ✅ | 1M tokens | Free |
Please note that some features may not be available within Giselle even if they are offered in the official API.
gemini-2.5-pro-exp-03-25
The latest comprehensive model supporting all features, with state-of-the-art capabilities. It offers text generation, web search, advanced reasoning, and robust coding features. It supports a wide range of input formats including PDF, images, audio, and video, and includes a context window of up to 1 million tokens (with expansion to 2 million tokens planned). Ideal for handling complex problems, deep multimodal understanding, and tasks requiring extensive data analysis.
gemini-2.0-flash-001
Optimized for speed, this model provides fast text generation and efficient web search functionality with support for PDF, image, audio, and video inputs, but lacks complex reasoning capabilities. With a context window of 1 million tokens, it’s best suited for scenarios requiring quick, responsive interactions and moderate multimodal input handling.
gemini-2.0-flash-thinking-exp-01-21
A reasoning-focused, rapid-response model featuring advanced logical reasoning capabilities and multimodal support (PDF, images, audio, and video). This model does not include web search functionality but is optimized for tasks requiring analytical depth, critical thinking, and efficient multimodal reasoning within a context window of 1 million tokens.
gemini-2.0-pro-exp-02-05
A balanced, professional-level model offering robust text generation and effective web search capabilities. Supports multiple inputs such as PDF, image, audio, and video but excludes advanced reasoning capabilities. With a context window of 1 million tokens, it is well-suited for general multimodal generation and standard use cases.
Model Selection Guide
Guidelines for selecting the optimal model:
- For best overall performance and complex tasks: gemini-2.5-pro-exp-03-25
- When speed and efficiency are a priority: gemini-2.0-flash-001
- For advanced reasoning and logical analysis: gemini-2.0-flash-thinking-exp-01-21
- For balanced general-purpose performance: gemini-2.0-pro-exp-02-05
Practices for Giselle
We recommend gemini-2.5-pro-exp-03-25 as your primary model in Giselle. It’s the most powerful option available, capable of processing large volumes of data, executing complex code reviews, and handling a wide variety of business tasks. By enabling the Search grounding feature, you can access web search functionality, allowing you to supplement your workflows with the most current information available.
For detailed specifications, performance benchmarks, or additional assistance, please check Google AI for Developers.