Google has unveiled Gemini 2.5 Pro, a new AI model with improved thinking and reasoning capabilities. This update promises significant performance improvements in areas such as mathematics, programming and multimodal processing.
Key features:
- Advanced thinking processes: The model processes tasks step-by-step and provides more precise answers to complex questions.
The Verge - Superior benchmark results: Gemini 2.5 Pro leads common AI benchmarks and outperforms competitors such as OpenAI and Anthropic.
The Verge - Multimodal capabilities: Supports interpretation of text, audio, images, video and code.
- Large context window: Currently 1 million tokens, with a planned expansion to 2 million tokens.
Improved thinking and reasoning skills
Gemini 2.5 Pro is characterized by its ability to analyze information, draw logical conclusions and make informed decisions. These improvements result from the integration of advanced techniques such as reinforcement learning and chain-of-thought prompting. In benchmark tests such as “Humanity’s Last Exam”, the model achieved a remarkable score of 18.8% without any additional tools.
Outstanding performance in benchmarks
Gemini 2.5 Pro has proven its performance in various tests.
- GPQA Diamond: scored 84%, demonstrating a deep understanding of complex problems.
- AIME 2025: With 86.7 %, the model shows its strength in mathematical and scientific areas.
- SWE-Bench Verified: Scored 63.8%, demonstrating its ability to solve real-world software problems.
Multimodality and extended contextual understanding
An outstanding feature of Gemini 2.5 Pro is its native multimodality. The model can interpret not only text, but also audio, images, videos and code. With a current context window of 1 million tokens, which will soon be expanded to 2 million, it is able to process extensive data sets and solve complex problems.
Availability and outlook
Gemini 2.5 Pro is now available for Google AI Studio users and Gemini Advanced subscribers. In the coming weeks, the model will also be integrated into Vertex AI. Google also plans to introduce pricing models in the near future to enable scaled usage.
With Gemini 2.5 Pro, Google is setting new standards in AI development and offering a powerful tool for developers and companies who want to solve complex tasks efficiently.
