Google (GOOG, Financial) introduced Gemini 2.5 on Tuesday, its latest large language model designed to bring advanced reasoning capabilities to artificial intelligence applications.
The company described Gemini 2.5 as a “thinking model” that improves response accuracy by processing information more deeply before answering. According to a company blog post, the model analyzes data, applies context, draws logical conclusions, and makes decisions—key components of what it defines as reasoning in AI.
Gemini 2.5 is built on an upgraded base model combined with refined post-training, which Google said allows for better performance and supports the development of more capable and context-aware AI agents.
The launch includes Gemini 2.5 Pro Experimental, described by Google as its most advanced model for complex, multimodal tasks. The company said it outperforms comparable models, including OpenAI's o3-mini and GPT-4.5, Claude's Sonnet 3.7, Grok 3 Beta, and DeepSeek's R1.
Gemini 2.5 Pro Experimental is currently accessible through Google's AI Studio, the Gemini app for Advanced plan subscribers, and is expected to arrive on Vertex AI soon. Pricing details will be provided in the coming weeks.
