Google Launches Gemini 2.5 Pro: Powerful Reasoning, Dubbed "the Smartest AI Model" 164

Google released its latest ai model Gemini 2.5 Pro on March 25, 2025, which is currently its "most intelligent" model. It features powerful reasoning capabilities and multimodal support. It has performed exceptionally well on multiple benchmarks, particularly excelling at complex task processing and code generation.

Gemini 2.5 Pro Model Features

• Reasoning Capabilities: Gemini 2.5 Pro is a "thinking model" capable of processing tasks through reasoning to provide more accurate answers. It has performed exceptionally well on multiple benchmarks—for example, on Humanity's Last Exam (HLE), the model achieved a score of 18.8% without using tools, surpassing other competitors.

• Multimodal Support: The model supports multimodal input and can process text, audio, images, video, and code repositories.

• Context Window: Gemini 2.5 Pro currently offers a context window of 1 million tokens, which will soon be expanded to 2 million tokens.

Gemini 2.5 Pro Coding Capabilities

Gemini 2.5 Pro excels at coding, capable of generating complex code from simple text prompts. For example, it can generate a simple endless runner game within a minute. In addition, the model also performs exceptionally well in creating visually appealing web applications and agentic code applications.

Gemini 2.5 Pro Performance

Benchmarks: Gemini 2.5 Pro ranks at the top across multiple benchmarks—for example, achieving 84% on GPQA Diamond and 86.7% on AIME 2025.

LMArena Ranking: The model ranks first on the LMArena leaderboard, surpassing Grok 3 and GPT-4.5.

Gemini 2.5 Pro is currently only available to Gemini Advanced users, who can use the model in Google AI Studio and the Gemini application. It will be expanded to Vertex AI in the coming weeks.

As for pricing, Google has not yet announced it, but has confirmed that it will be released soon.

Usage address: aistudio.google.com