Skip to content
#AI

Gemini 2.5: Our most intelligent AI model (AI column)

#AI #Software #ML

Interesting. announcement From Google, where they announced their coolest model to date. Turns out she's SOTA. (state of the art) On a large number of benches and bypasses models from Open AI and other players. That's what she does.

  • Reasonability: Gemini models 2.5 They have built-in “thinking” capabilities that allow them to analyze information and draw logical conclusions before giving a response. This model is similar to Claude Sonnet. 3.7
  • Improved performance on benches: Gemini 2.5 The Pro outperformed previous versions and competing models in various tests including LMArena, AIME 2025 and GPQA diamond.
  • Expanded contextThe model supports the context window to 1 A million tokens, further promised to increase it to 2 millions.
  • Multimodality: Gemini 2.5 can process various types of data, including text, audio, images, video and code.
  • Accessibility: Experimental version of Gemini 2.5 Pro is already available in Google AI Studio annex Gemini For Gemini Advanced subscribers.

Anyway, it looks like the guys at Google were able to get together and give a cool result.

#AI #Software #ML