Gemini 2.5 Pro Is Google’s Most Powerful AI Model, and It’s Already Free
Google has released a shiny new AI model in the form of Gemini 2.5 Pro, albeit with an experimental label next to it – and it’s available for free, so you don’t need to sign up for Gemini Advanced to get it. As with many recent releases of artificial intelligence models , the model’s “reasoning” capabilities are considered the biggest update.
In artificial intelligence terms, reasoning means more carefully crafted responses. According to Google, this should lead to fewer errors, more logical answers, and a better understanding of “context and nuance.” This extra “thinking” ability will now come standard in future Google models.
The Pro (experimental) edition is the first variant of Gemini 2.5 to appear, and while the original blog post didn’t mention free users, less than a week later we received an update that says it’s available for everyone – with speed limits applying if you’re not a Gemini Advanced subscriber (Google hasn’t specified what those speed limits are). The new model is already available through the desktop application and will soon appear on mobile devices.
Google points to several benchmark tests that show the capabilities of the Gemini 2.5 Pro. At the time of writing, it topped the LMArena leaderboard , where users rate responses from dozens of artificially intelligent chatbots. It also scored 18.8 percent on the Humanity’s Final Exam test, which measures human knowledge and thinking, slightly ahead of rival models OpenAI and Anthropic.
Also worth noting is the large context window. Simply put, it’s a measure of how much data an AI model can process at one time, and Gemini 2.5 Pro has a context window of one million tokens, with two million “coming soon,” according to Google. This is comparable to a context window of, for example, 200,000 tokens for the o3-mini ChatGPT reasoning model.
As is typical with AI claims, there is no mention of copyright infringement regarding training data or increased power consumption. Current artificial intelligence models use “staggering” amounts of electricity and water and have put us on an “unsustainable path” that needs to be reversed quickly, according to researchers at MIT .
Gemini 2.5 Pro test
Quantifying improvements from one AI model to another can be difficult, which is why benchmarks like LMArena are useful. I lack the expert scientific knowledge or programming knowledge required to really test the Gemini 2.5 Pro, although, as with the previous model, I was able to create a few simple web applications (like an online timer) in a matter of minutes.
I know a little about Bleak House by Charles Dickens , so I fired up Gemini 2.5 Pro to work on the text. He gave me an accurate summary of the plot and an intelligent assessment of the various narrative techniques used (which would have helped me a lot during my studies). It also turned the book into a fairly well-made three-act movie structure – evidence that he holds a lot in his “mind” at once.
The older version of Gemini 2.0 Flash could also accurately answer the same Bleak House clues, but Gemini 2.5 Pro’s answers were longer, more detailed, less general, and smarter – evidence that this extra “reasoning” was involved. The Gemini 2.0 Flash model also had to split the film adaptation into three responses, possibly due to the sheer amount of text it was trying to process.
Google has provided its own example of Gemini 2.5 Pro’s capabilities, showing how you can create a simple endless runner with just one hint. Although the demo video showing the code output is sped up, the game appears to work and is quite well designed, which is an impressive end result when using a single natural language hint. There’s also a nifty web demo of a digital fish swimming.
Online, a new artificial intelligence model is being thoroughly tested. Software engineer and independent AI researcher Simon Willison ran several tests covering image generation, audio transcription and code generation, and was very pleased with what Gemini 2.5 Pro could come up with.
The frantic pace of artificial intelligence development shows no signs of slowing down anytime soon, and we can expect new Gemini 2.5 models to appear in the near future. “As always, we welcome feedback so we can continue to improve Gemini’s impressive new capabilities at a rapid pace, all with the goal of making our AI more useful,” says Koray Kavukcuoglu of Google’s DeepMind AI Lab.