Support our independent tech coverage. Chrome Unboxed is written by real people, for real people—not search algorithms. Join Chrome Unboxed Plus for just $2 a month to get an ad-free experience, access to our private Discord, and more. Learn more about membership here.
START FREE TRIAL (MONTHLY)START FREE TRIAL (ANNUAL)
Google is moving fast with its AI development, and its most intelligent model, Gemini 2.5 Pro, is getting one last significant tune-up before it’s officially ready for primetime. Google just recently announced an upgraded preview of the model (newer than the one they debuted last month), noting that this improved version is the one that will become the generally available, stable release in just a couple of weeks, ready for large-scale enterprise applications. That means the best version of Gemini is getting even better, and we’ll all be benefiting from these improvements soon.
Smarter, more creative, and better at coding
So, what’s new in this upgraded preview? It’s not just about minor tweaks; Google is touting some serious performance gains and qualitative improvements based on user feedback from the version released in May.
On the performance front, Gemini 2.5 Pro has seen a significant jump on key industry leaderboards. It increased its score on the LMArena (which evaluates human preference for model responses) and the WebDevArena (a popular coding leaderboard), maintaining its leading position on both. Google also notes that it continues to excel at difficult coding benchmarks and shows top-tier performance on highly challenging math, science, and reasoning tests.
Perhaps more importantly for everyday users, Google says it specifically addressed feedback about the model’s output. This new version has an improved style and structure, meaning it can be more creative and provide better-formatted responses. This is a welcome refinement, as getting a well-organized and clearly presented answer can be just as important as the factual accuracy of the information itself.
More control for developers, better experiences for users
For developers building with the model, Google has extended “thinking budgets” to this new version of 2.5 Pro. This is a feature that gives developers more control over the model’s cost and latency by letting them adjust how much processing “thought” the model puts in before responding. While a developer-focused feature, this ultimately benefits end-users, as it can lead to faster and more efficient AI-powered apps and services.
How to try the upgraded Gemini 2.5 Pro
The best part is, you can experience these improvements right now. The upgraded preview of Gemini 2.5 Pro is available today for developers in the Gemini API via Google AI Studio and Vertex AI. For everyone else, Google has stated that it is also rolling out now in the Gemini app, so you can directly interact with this more polished and powerful version of Google’s best AI.
SUBSCRIBE TO UPSTREAM
Get Chrome Unboxed delivered straight to your inbox
Upstream is our flagship, curated newsletter with the top stories, most click-worthy deals, giveaways, and trending articles from Chrome Unboxed sent directly to your inbox a few times a week. Join 31,000+ subscribers.
