Google said in a blog post on Tuesday that Gemini 2.5 is a new AI inference model that will compete with DeepSeek R1 and is currently the most watched AI model on LMARENA.
Google describes the new series of Gemini 2.5 models as “thinking models” that recursively analyze answers before giving people the final output. With every benchmark on LMARENA, Gemini 2.5 leads in reasoning, science, mathematics, and proxy coding. However, it did not win in all tests. For example, OpenAi O3-Mini is still booting it in LiveCodeBench V5.
Gemini 2.5 is now rolling out to paid premium users. Some users on Reddit reported that they needed to delete and reinstall the Gemini app to show 2.5. On the desktop, Gemini 2.5 is available on Google AI Studio.
The advantage Google’s AI model has over competitors is its high high token rate – the ability to understand or generate complex data sets. Google has been touting Gemini as an AI that can handle large context windows with high token output. On X, the social media platform formerly known as Twitter, people are still trying out the features of Gemini 2.5. Fei Xia is an employee researcher at Google DeepMind, who is able to roughly draw three-layer cakes and convert them into 3D printed files.
Gemini 2.5 Pro is such a powerful coding model. I was able to use it to convert the image into a 3D printed object and implement it. More pic.twitter.com/6whk3nydwd
– fei Xia (@xf1280) March 25, 2025
Google has released a video showing Gemini 2.5 making a simple endless runner video game in seconds. Another guy on X made a simple flight simulation video game.
Google did not immediately respond to a request for comment.
The launch of Gemini 2.5 is to put the latest weapons into the AI gladiator. Earlier this year, DeepSeek R1 launched the US AI company. DeepSeek has released a free open source inference model that is more effective than Openai offers.
Google is remanded on AI. From search to documents, this generation technology has penetrated almost everything in the company’s product portfolio. Google plans to invest $75 billion in AI development in 2025 alone. Given that the AI market is expected to grow to $1.8 trillion by 2030, Google has a strong financial motivation to occupy that space, according to Grand View Research.
In addition to Gemini 2.5, Google also launched Gemini 2.0 Flash Thinking earlier this year, an idea designed to be a faster model of reasoning. Last month, Google released Gemini Code Assist, a free AI encoding tool with spacious input token support.
