Google is updating Gemini with a fresh release of Gemini 2.5 Flash. This update, now available in the Gemini app, makes the AI’s responses clearer, more detailed, and better at handling visuals.
A major change is how answers are organized. Gemini now uses headers, bullet points, and tables to present information more clearly. If you ask a complex question, the app breaks it down so it’s easier to follow.
Gemini is also improving how it works with images. You can upload a diagram, handwritten notes, or study material, and the model can analyze it, summarize key points, or turn it into flashcards. This is especially useful for students and professionals who use visual materials.
Google is also releasing developer previews of Gemini 2.5 Flash and a lighter Flash-Lite version in AI Studio and Vertex AI. These previews let developers try out new features and share feedback before the full release.
Double the token efficiency, faster replies
Credit: Google
A key improvement is efficiency. Gemini 2.5 Flash now uses tokens twice as efficiently as before, so the same content needs fewer tokens. This means lower latency, faster replies, and possibly lower costs for businesses. In tests where the model had to connect different steps or tools, Google saw about a 5% boost in performance, which helps with reliability.
Flash-Lite is also getting updates. This lighter version now uses fewer words, follows complex instructions better, and has improved at tasks like translation, image analysis, and audio transcription.
Google has not said when these updates will be available everywhere, but the developer previews show that more changes are coming before the full launch.
