Google just dropped Gemini 2.5, a new generation of AI models focused on advanced reasoning—starting with the launch of Gemini 2.5 Pro Experimental, now ranked as the company’s smartest model to date.
Key Highlights:
#1 on LMArena: Gemini 2.5 Pro debuts at the top of the leaderboard, showing state-of-the-art reasoning across math, science, and coding.
Strong Coding Performance:
Scores 63.8% on SWE-Bench Verified
Scores 68.6% on Aider Polyglot, with notable strengths in web app dev and agentic code workflows
Massive Context Window:
Ships with a 1 million token context window—Google says this will soon expand to 2 million, allowing full ingestion of codebases and large datasets.
Availability:
You can try Gemini 2.5 Pro today in Google AI Studio and the Gemini app (Advanced tier). API pricing is expected in the coming weeks.
Why it matters:
Google is turning reasoning into a baseline feature, not a premium one. While it doesn’t generate as much buzz as OpenAI, its pace in pushing state-of-the-art models remains relentless. But with GPT-5 and others on the horizon, how long Gemini 2.5 holds the top spot is anyone’s guess.