Google on Wednesday morning ET launched Gemini 2.5 Pro, its most intelligent model yet.
The model is launched at an experimental level, as is usual for Google and available for free in the Google AI Studio for developers.
Google in the launch promotions highlighted the model has outperformed top models from rivals like OpenAI’s o3-mini and GPT-4.5 and Anthropic’s Claude 3.7 Sonnet and Elon Musk-led xAI Grok 3 Beta on key benchmarks.
Among third-party reviews, the new model has climbed to the top of LMArena’s Chatbot leaderboard, surpassing Grok-3-Preview and GPT-4.5-Preview.
It has also surpassed OpenAI’s o3-mini-high on AIME 2025 benchmark at 86.7% against the latter’s 86.5%.
Gemini 2.5 Pro retains the long context window that got the line of models a chance at fame at 1 million, with plans to expand to 2 million.
Google highlighted how the new model can perform complete programming tasks like creating a game from scratch — a task that competitor models like Claude 3.7 Sonnet also excel at performing.
The model is also capable of reasoning before it responds to plan out the response and share it with the user.
Google CEO Sundar Pichai dubbed the launch a “start of a new era of thinking models.”
“We can’t wait to see where things go from here,” Pichai said.
The company timed its launch to be just an hour ahead of when OpenAI was set to make an undisclosed announcement at 11 am pacific time.
Planning your financial journey can be daunting but it doesn't have to be. Fire Fast by Dzambhala helps you understand and plan effectively.
Join the vibrant privacy-ensured Dzambhala community on
Want to give feedback on this story? Write to us.