Overview

• The update boosts Gemini 2.5 Pro’s coding performance above competitors, delivering a 24-point Elo gain on LMArena and topping benchmarks like Aider Polyglot
• Reasoning abilities have been enhanced to handle complex math, science and coding tasks through multi-step analytical processing
• Paid subscribers can now submit up to 100 prompts per day, doubling the previous limit as demand and confidence in the model grow
• Feedback-driven improvements deliver more creative and better-formatted responses after addressing style and structure issues
• Google intends to promote this build to a long-term stable release in the coming weeks and roll it out across its AI services