Gemini 2.5: Meet Google’s Most Intelligent Reasoning Model To Date

It’s only been a few months since the debut of Gemini 2.0. But at the breakneck pace Google is moving at with AI development, December probably seems like ancient history. The company is already moving on to the AI model’s next update — Gemini 2.5.

Today, Google announced Gemini 2.5, its latest and most intelligent AI model to date. The first release of the 2.5 generation will be Gemini 2.5 Pro, which is a thinking model (capable of reasoning through thoughts before responding) just like Gemini 2.0 Flash Thinking. Pro will come with a 1 million token context window, and Google says this will expand to 2 million soon.

According to the tech giant, Gemini 2.5 outperforms some of the most popular AI models across a range of benchmarks, including reasoning, science, and mathematics. The company achieved this feat by “combining a significantly enhanced base model with improved post-training.”

It appears that enhancing coding performance was a major focus this time around. The company claims it made a big improvement compared to 2.0, allowing it to create “visually compelling web apps and agentic code applications.” To show off the model’s coding capabilities, Google shared this video of the AI generating a game based on a text prompt.

Gemini 2.5 is launching as an experimental model that will be available to developers in AI Studio and Gemini Advanced users in the Gemini app. If you’re a Gemini Advanced user, you’ll be able to select Gemini 2.5 from the model dropdown on desktop and in mobile app. Google says Gemini 2.5 will also be available for Vertex AI in the coming weeks, but it did not offer an exact date.

Got a tip? Talk to us! Email our staff at [email protected]. You can stay anonymous or get credit for the info, it’s your choice.