Google Unveils Gemini 2.5 Pro, Its Most Advanced AI Model Yet

Google has introduced Gemini 2.5, its most sophisticated artificial intelligence model, incorporating step-by-step reasoning to enhance decision-making. This update significantly improves the model's ability to interpret complex queries and generate precise, context-aware responses.

Google describes Gemini 2.5 Pro as an experimental model built for intricate tasks, achieving top rankings on the LMArena leaderboard, which evaluates human preference. The model demonstrates superior reasoning and coding performance, excelling in coding, mathematics, and science benchmarks.

All Gemini 2.5 models are designed to process information methodically, ensuring greater accuracy and reliability. Google emphasizes integrating these capabilities into all AI models to enhance problem-solving and create more context-aware AI agents.

A major improvement in Gemini 2.5 Pro is its coding proficiency, marking a significant leap from its predecessor. It delivers enhanced capabilities in web app development, agentic coding applications, and code editing. The model scores 63.8 per cent in SWE-Bench Verified evaluations, reinforcing its strengths in software engineering.

Gemini 2.5 Pro is now accessible in Google AI Studio and the Gemini app for advanced users. Google plans to integrate the model into Vertex AI and introduce pricing details for enterprise-scale deployment in the coming weeks.

With expanded multimodal support, Gemini 2.5 Pro currently features a 1 million token context window, set to increase to 2 million tokens soon.