Google has released Gemini 2.5, an AI language model with improved reasoning, coding, and multimodal capabilities. The model outperforms previous versions and achieves cutting-edge results on AI benchmarks, excelling in code transformation and logical reasoning.
Gemini 2.5, an updated AI language model with better reasoning, coding, and multimodal capabilities, was released by Google. The new model, known as Gemini 2.5 Pro Experimental, outperforms earlier iterations and exhibits cutting-edge results on a range of AI benchmarks. Take a look at Sundar Pichai's 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
The coding capabilities of Gemini 2.5 are significantly better than those of Gemini 2.0. The model performs exceptionally well in tasks involving code transformation, online application development, and agile code applications. On SWE-Bench Verified, a benchmark for assessing coding agents, it receives a score of 63.8%. Google also presented an example in which the model used a single-line prompt to generate executable code for a video game.
Also Read | Apple WWDC 2025 dates announced: Here's what you can expect in June
Gemini 2.5's enhanced reasoning capability is one of its main improvements. The model is more effective than its predecessor in analysing data, taking context into account, and coming to logical conclusions. Google credits improved post-training methods and an improved base model for this improvement. With a score of 18.8% on Humanity's Last Exam, a dataset used to evaluate AI's capacity to manage challenging knowledge-based questions, the model surpasses criteria in both science and mathematics.
Google AI Studio and the Gemini app for Gemini Advanced customers now provide Gemini 2.5 Pro. In the upcoming weeks, it will be available through Vertex AI, and further price options will be unveiled soon. In order to stay competitive in AI-driven applications, Google intends to incorporate sophisticated reasoning skills into all next Gemini models.