Google debuts Gemini 2.5

Google DeepMind introduced Gemini 2.5, describing it as its most intelligent “thinking” model to date. The first release, Gemini 2.5 Pro (Experimental), debuts at #1 on the LMArena human-preference leaderboard and leads common coding, math and science benchmarks, including GPQA and AIME 2025. It records 18.8% on Humanity’s Last Exam (no tools) and 63.8% on SWE-Bench Verified with a custom agent setup. The model ships with native multimodality and a 1-million-token context window, with a 2-million window “coming soon,” aimed at handling large datasets across text, audio, images, video and code. 2.5 Pro is available in Google AI Studio and in the Gemini app for Gemini Advanced users; availability on Vertex AI is slated to follow, with pricing to be announced. DeepMind says “thinking” capabilities—structured internal reasoning before responding—will be built into all future Gemini models to support more capable, context-aware agents.