New releases are coming in thick and fast in the AI space.
After DeepSeek had released a new version of its V3 non-reasoning model earlier in the day, Google has released its own Gemini 2.5 Pro model. Google says the reasoning model is the company’s strongest model yet. Google shared data showing that the model was beating publicly-available models by companies including OpenAI, xAI, Anthropic and Deepseek.
“Gemini 2.5 is here, and it’s our most intelligent AI model ever,” Google CEO Sundar Pichai posted on X. “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 LM Areana by a significant margin. With a model this intelligent, we wanted to get it to people as quickly as possible. Find it on Google AI Studio and in the @geminiapp for Gemini Advanced users now – and in Vertex in the coming weeks. This is the start of a new era of thinking models – and we can’t wait to see where things go from here,” he added.
Google also shared how Gemini 2.5 Pro compared to competitors in its blog. On benchmarks ranging from Humanity’s last exam, to AIME 2024, to Livecodebench and Aider Polyglot, Gemini 2.5 Pro compared very favourably against top models from OpenAI, xAI, Anthropic and DeepSeek.
In particular, Gemini 2.5 Pro scored the highest ever score of 18.8 percent on the Humanity’s last exam, compared to the 14 percent that o3-mini had managed. In the science focused GPQA diamond benchmark, it scored 84 percent, higher than the 79.7 percent OpenAI’s o-3 mini had managed. In math’s AIME 2025 benchmark too, Gemini 2.5 Pro was marginally ahead of o3-mini high.
“With Gemini 2.5, we’ve achieved a new level of performance by combining a significantly enhanced base model with improved post-training. Going forward, we’re building these thinking capabilities directly into all of our models, so they can handle more complex problems and support even more capable, context-aware agents,” Google said in its blog.
Google had been caught napping when OpenAI had released ChatGPT in November 2022. For the next couple of years, Google had largely trailed OpenAI on most benchmarks. But in 2025, Google seems to have hit its stride, and is now beating OpenAI on many benchmarks. It remains to be seen how OpenAI responds, but Google finally seems to be coming into its own in the AI race.