Gemini 2.5 Pro: update marks a new level in Google's artificial intelligence

Inputs and outputs

• Input options : Text, code, images (up to 3,000), audio (up to 4 hours), video (up to 10 files, 1 hour), documents (up to 3,000 PDFs)
• Outputs : Structured text

Limits

• Input : up to 1,048,576 tokens
• Standard output : up to 65,535 tokens
• Supported input size : up to 500 MB

Available resources

• Multimodal reasoning with context caching
• Based on information obtained via Google Search
• Code execution
• System instructions, function calls
• Support for the Vertex AI RAG engine
• Batch forecasting and token counting

Resources not available

• Fine-tuning
• Live API preview (available only in the Pro version)

Generation

• Temperature : 0.0 to 2.0 (default: 1.0)
• topP : 0.0 to 1.0 (default: 0.95)
• topK : fixed at 64
• Candidate count : 1 to 8

Available regions

• North America : us-central1, us-east1, us-east5, us-south1, us-west1
• Europe : europe-central2, europe-north1, europe-west1, europe-west4, europe-west8

Gemini 2.5 Pro performance at WebDev Arena – Image: Google for Developers

Gemini CLI: Google's command-line coding assistant

While Gemini 2.5 Pro focuses on enterprise, productivity, and multimodal applications, another recent Google release shows how AI can also accelerate the work of developers: the Gemini CLI .

This is an open-source AI agent designed to provide direct support in the programming terminal. With it, you can generate code snippets, document scripts, debug errors, and even execute commands autonomously using natural language.

Official icon for Gemini CLI, Google's new coding assistant – Image from the Google Blog

The Gemini CLI can be installed via the terminal and connects to the Google API to interpret commands and deliver contextualized solutions, reducing development time and making coding more accessible.

The connection between Gemini 2.5 Pro and Gemini CLI reinforces Google's strategy of creating an interoperable AI ecosystem, encompassing everything from complex enterprise solutions to the everyday work of developers.

Also read: “Google VEO 3: the future of AI-powered videos in 2025”

Practical applications of Gemini 2.5 Pro

The new model stands out in contexts such as:

• Corporate productivity : automating tasks and delivering summaries at scale.
• Analysis and decision-making : with contextualized responses based on structured data.
• Content creation and prototyping : texts, creative ideas, and real-time suggestions.

Personalization and real-time processing make the model especially useful for companies looking to scale their AI operations with less technical effort involved.

For developers: Vertex AI, API, and integration with Google AI Studio

Gemini 2.5 Pro can be easily used through Google AI Studio for rapid prototyping, and robustly integrated via API for customized solutions. The official documentation provides a detailed implementation guide.

Furthermore, Vertex AI allows for managing models, running batch inferences, and scaling safely and efficiently.

Conclusion: a broader, more accessible, and more powerful AI

The launch of Gemini 2.5 Pro consolidates Google's commitment to a more multimodal, integrated, and capable . In parallel, the Gemini CLI shows that this advancement is also reaching the developers' terminals.

The expectation is that, together, these resources will define a new standard for productivity and digital intelligence in the coming years. We are increasingly approaching a reality where AI becomes not just a tool, but an essential infrastructure for innovation .

Written by Skyone