Google unveils Gemini 2.5, its most advanced AI model yet

Google has taken the wraps of Gemini 2.5, its most advanced artificial intelligence model so far. This represents a huge leap in the world of AI as the latest version incorporates “thinking” and “reasoning” capabilities, bringing it closer to human-like intelligence. The model can process tasks step-by-step and make decisions based on reasoning.

Naturally, Gemini 2.5 will be able to handle far more complex tasks than previously possible thanks to its advanced thinking capabilities. The model will be able to provide users with more contextually accurate and relevant responses.

Google Gemini 2.5: What can it do better?

Google’s Gemini 2.5 models, including all future versions, are built as ‘thinking models’ that are capable of reasoning before responding. The models can tackle complex prompts and generally, be more contextually aware, therefore, providing more accurate responses.

ALSO READ: Hey Gemini, what’s this?: Google’s AI just got a lot more human

Google states that reasoning abilities are now embedded directly into all its models. Say you want to open a new cafe in City A versus City B, it could give you the relevant pros and cons based on your business model and could even reason the best course of action by using the latest data about both cities.

What is Gemini 2.5 Pro and who can use it?

Gemini 2.5 Pro, codenamed Nebula, is the first model in the Gemini 2.5 series, and it excels at performing complex tasks. Gemini 2.5 Pro (experimental) is now available for Gemini Advanced subscribers and Google AI Studio developers. Google has stated that Gemini 2.5 Pro “tops the LMArena leaderboard, which measures human preferences, by a significant margin”.

ALSO READ: Gemini vs ChatGPT: What AI chatbot should you choose?

It also leads on math and science benchmarks like AIME 2025 and GPQA diamond. Gemini 2.5 Pro is also tailor-made to create visually appealing web pages and agentic code applications. In addition to native multimodality, Gemini 2.5 Pro now has a 1 million token context window with another 2 million coming soon.

To put it simply, this means it can comprehend and handle vast datasets and complex problems more easily. Be it from text, audio, images, video, or code, the 2.5 Pro model can comprehend and reason better.

Gemini 2.5 Pro replaces Gemini 2.0 Pro (experimental) in the Gemini app, so if you have an Advanced subscription, you can check it out. To know more about the latest in the world of tech, stay tuned to Unboxed by Croma.