Google has announced Gemini 2.5, its latest AI reasoning model designed to enhance problem-solving and fact-checking capabilities. The model, which is available in Google AI Studio and the Gemini app for $20-a-month Gemini Advanced subscribers, represents the company’s most advanced AI to date. Moving forward, all new Google AI models will incorporate reasoning capabilities.
Since OpenAI introduced its first AI reasoning model in 2024, competitors like Anthropic, DeepSeek, and xAI have worked to develop their own versions. These models use additional computing power to analyze problems before responding, improving accuracy in tasks like coding and mathematics. However, they are also more expensive to run.
Gemini 2.5 is built to excel in visually compelling web apps and agentic coding applications. Google claims it outperforms its previous frontier AI models and some top competitors. On the Aider Polyglot benchmark for code editing, it achieved a 68.6% score, surpassing OpenAI, Anthropic, and DeepSeek. On the SWE-bench Verified test for software development, it scored 63.8%, outperforming OpenAI’s o3-mini and DeepSeek’s R1 but falling short of Anthropic’s Claude 3.7 Sonnet, which scored 70.3%.
Another key benchmark, Humanity’s Last Exam, evaluates AI performance across math, humanities, and natural sciences. Gemini 2.5 Pro scored 18.8%, surpassing most flagship AI models. It also features a one-million-token context window, allowing it to process up to 750,000 words at once—longer than the entire Lord of the Rings book series. Google plans to expand this to two million tokens soon.
While Google has not yet disclosed API pricing for Gemini 2.5 Pro, it has promised to release details in the coming weeks.
