Google Unveils Latest Addition to the Gemini 2.5 Family

Google has introduced its latest model, Gemini 2.5 Flash-Lite, available in preview, alongside the general release of the more potent 2.5 Pro and Flash models. Flash-Lite surpasses its 2.0 predecessors in coding, math, and reasoning, offering a cost-effective option at $0.10 per million input tokens and $0.40 per million output tokens. However, its baseline performance is limited, scoring just 5.1% on Humanity’s Last Exam. A standout feature is its “Thinking” mode, which enhances processing time and dramatically improves performance on challenging tasks, like increasing its score on the AIME 2025 math benchmark from 49.8% to 63.1%. Developers can begin experimenting with Flash-Lite immediately via Google AI Studio and Vertex AI. The more advanced Flash and Pro models are also available there, integrated into the main Gemini app for consumers, and custom versions are already in use to support Google’s Search engine.

Source link