Gemini 2.5 Pro Update: Enhanced Coding, Math, and Instruction-Following Capabilities
Google has rolled out a significant update to its Gemini 2.5 Pro model, announced on June 5, 2025, positioning it as a formidable contender in the AI landscape.
The latest Gemini 2.5 Pro excels in coding tasks, leading on challenging benchmarks like Aider Polyglot with a score of 86.2%, compared to o3’s 81.3%.
Its ability to handle complex coding projects has been enhanced, making it a top choice for developers building interactive web applications.
In mathematics, the model demonstrates superior performance on benchmarks such as GPQA and AIME 2025, with a notable 24-point Elo score jump on LMArena, reaching 1470, and a 35-point increase on WebDevArena, hitting 1443.
It also leads on Humanity’s Last Exam (HLE), a rigorous test of reasoning and knowledge, further solidifying its edge in math and science domains.
Developers can also leverage new features like thinking budgets, which allow fine-tuned control over cost and latency, making it more practical for enterprise-scale use.
Also read:
- DeepSeek R1 Soars to Top 3, Outshining Gemini—An Open-Source Triumph in Coding, Math, and Logic
- The True Brake on Civilization: Coordination Limits Are Rapidly Shifting
- "Ballerina" Delivers Stunning Action in a Thrilling "John Wick" Spin-Off
While currently in preview in Google AI Studio, the updated Gemini 2.5 Pro is set to become generally available in the Gemini API and Vertex AI within the coming weeks, promising stable production use.
This release not only highlights Google’s commitment to advancing AI but also raises the bar for what developers can achieve with open-access tools, offering a competitive alternative to proprietary models like o3. As Gemini 2.5 Pro prepares for broader deployment, its enhanced capabilities are poised to accelerate innovation across industries.