Google Supercharges Gemini 2.5 With Deep Think and Smarter Flash

May 20, 2025 — Google just upgraded Gemini 2.5 Pro and Flash, with big wins for performance, reasoning, and developer tools.

Google’s Gemini 2.5 Pro now tops coding and learning leaderboards like WebDev Arena and LMArena. It handles long context and video better than ever, thanks to a 1 million-token window. Deep Think, its new reasoning mode, pushes performance further. It scored high on 2025 USAMO and outpaced rivals in coding and multimodal tests.

Flash, the lightweight model, now runs faster and uses fewer tokens—down 20–30%—while improving in reasoning, long context, and code.

Also read: Gemini in Gmail Now Handles Google Calendar Tasks on Android and iOS

Developers get new features across the board. Pro and Flash now support thought summaries and thinking budgets in the API and Vertex AI. Pro gains support for MCP tools and tighter control over token use. Both models now preview native audio output, multi-speaker text-to-speech, and affective dialogue. You can even direct tone and accent.

Google also added stronger defenses against indirect prompt injection and introduced Project Mariner’s computer use features into the API. Flash is already live in the Gemini app and Google AI Studio. The Pro update will follow in early June.

In other headlines, Google also recently held its I/O event with several new announcements. You can read about them here.

• Google I/O 2025: Top 10 Highlights of Google’s New Announcements
• Wear OS 6 Is Launched For Devs, Here’s What’s New in Google’s Smartwatch OS