Premium

Google’s Sundar Pichai calls Gemini 2.5 Pro Experimental a ‘state-of-the-art thinking model’

Gemini 2.5 Pro Experimental excels in reasoning, coding, and multimodal AI capabilities.

Gemini 2.5 Pro Experiment now available for advanced users. (Image credit: Google)

On Monday, Google unveiled Gemini 2.5 Pro Experimental, its latest, most capable, and also the first large language model from the Gemini 2.5 family. CEO Sundar Pichai calls it a “state-of-the-art thinking model,” which is now available for Gemini Advanced users and via Google AI Studio. It will be expanded to Vertex users in the coming weeks.

“Our first 2.5 model, Gemini 2.5 Pro Experimental, is a state-of-the-art thinking model, leading in a wide range of benchmarks—with impressive improvements in enhanced reasoning and coding—and now #1 on lmarena.ai by a significant margin. With a model this intelligent, we wanted to get it to people as quickly as possible,” wrote Sundar Pichai, CEO of Alphabet, Google’s parent company, in an X post.

1/ Gemini 2.5 is here, and it’s our most intelligent AI model ever.

Our first 2.5 model, Gemini 2.5 Pro Experimental is a state-of-the-art thinking model, leading in a wide range of benchmarks – with impressive improvements in enhanced reasoning and coding and now #1 on… pic.twitter.com/mtEdRCTcgF

— Sundar Pichai (@sundarpichai) March 25, 2025

lmarena.ai is an open platform for crowdsourced AI benchmarking, used to compare the performance of different large language models.

Also read | ChatGPT, powered by 4o, can now generate ‘beautiful and useful’ images

According to Google, Gemini 2.5 Pro Experimental outperforms models like OpenAI’s GPT-3o, Claude 3.7 Sonnet, Grok 3, and DeepSeek R1 in science and math test-based benchmarks such as GPQA Diamond and AIME 2025. It also excels in code editing AI benchmarks like Aider Polyglot.

Features and capabilities

Gemini 2.5 Pro Experimental is a multimodal AI model that can generate executable code for basic video games like Dino, the offline game many play on Google Chrome when experiencing internet issues.

It also boasts enhanced reasoning and advanced coding capabilities. Currently, it supports up to 1 million context tokens, with plans to expand to 2 million soon. Additionally, it can handle complex problems by integrating information from multiple sources, including text, audio, images, and video.