Google is rolling out a major update to its Gemini 2.5 Pro AI model, addressing concerns raised by users about performance outside of coding tasks. The new version boasts significant improvements in code generation, surpassing OpenAI and other models in the Aider Polyglot test.
According to Google, the updated Gemini 2.5 Pro will offer better formatting and creativity in responses, closing gaps in performance introduced with previous updates. Developers can now configure thinking budgets for more efficient use of the model.
The update aims to establish a stable release, eliminating the “Preview” disclaimer from the consumer-facing app and web interface. This version is expected to provide users with a smoother experience, solidifying Gemini 2.5 Pro’s position as a powerful AI tool in the market.
Source: https://arstechnica.com/ai/2025/06/google-releases-updated-gemini-2-5-pro-says-its-the-most-intelligent-model-yet