The Gemini family just got faster and smarter. Google has officially expanded its Gemini 3 lineup with the launch of Gemini 3 Flash, a model engineered to deliver frontier-level intelligence at unprecedented speed and a fraction of the cost. This release marks a significant step in making next-generation AI capabilities accessible to developers, enterprises, and everyday users across Google’s ecosystem.
Building on the strong foundation of Gemini 3 Pro and Gemini 3 Deep Think, Gemini 3 Flash combines Pro-grade reasoning with the low latency and efficiency the “Flash” series is known for. It’s designed not just for improved everyday tasks but is positioned as Google’s most capable model for complex, agentic workflows. Starting today, it’s rolling out globally:
* For Developers: Via the Gemini API in Google AI Studio, Gemini CLI, and the new agentic platform, Google Antigravity.
* For Everyone: As the default model in the Gemini app and AI Mode in Search.
* For Enterprises: Through Vertex AI and Gemini Enterprise.
Frontier Intelligence, Now at Scale
Gemini 3 Flash proves that speed doesn’t require a compromise on capability. It delivers standout performance on rigorous benchmarks, rivaling larger frontier models. It achieves impressive scores like 90.4% on GPQA Diamond and 81.2% on MMMU Pro, even outperforming its predecessor, Gemini 2.5 Pro, across several metrics.
Engineered for efficiency, it pushes the Pareto frontier of performance versus cost and speed. It intelligently modulates its “thinking” for complex tasks while using, on average, 30% fewer tokens than 2.5 Pro for everyday operations. The raw speed is a game-changer: it’s 3x faster than 2.5 Pro while being offered at a compelling price point of $0.50 per 1M input tokens and $3 per 1M output tokens.
For Developers: Intelligence That Keeps Pace
Built for iterative, high-frequency workflows, Gemini 3 Flash brings Pro-grade coding performance with minimal latency. It excels in agentic tasks, scoring 78% on the SWE-bench Verified benchmark—outperforming both the 2.5 series and Gemini 3 Pro. Its robust reasoning, tool use, and multimodal capabilities unlock new possibilities for developers:
- Creating intelligent in-game assistants or running A/B test experiments in near real-time.
- Performing complex video analysis, data extraction, and visual Q&A.
- Streamlining the design-to-code process, transforming static images into interactive experiences.
Leading companies like JetBrains, Bridgewater Associates, and Figma are already leveraging its speed and reasoning power to transform their business applications.
A Major Upgrade for Everyday Users
For the global user base, Gemini 3 Flash is now the default model in the free Gemini app, replacing 2.5 Flash. This upgrade supercharges daily interactions with AI through its incredible multimodal reasoning. Imagine:
* Analyzing a short video of your golf swing and receiving an instant improvement plan.
* Sketching a drawing and having Gemini guess what it is before you finish.
* Uploading an audio lecture to get a custom quiz that identifies knowledge gaps with detailed explanations.
* Dictating an app idea and watching Gemini build a functional prototype in minutes.
Furthermore, Gemini 3 Flash is enhancing AI Mode in Search globally. Building on Gemini 3 Pro’s reasoning, it better understands nuanced questions, pulling in real-time local data and web links to deliver comprehensive, visually organized answers at the speed of Search—perfect for planning last-minute trips or quickly grasping complex topics.
Ready to Experience the Speed?
Gemini 3 Flash is available now. Developers can start building with it in Google AI Studio, Google Antigravity, and Vertex AI. Everyone can access its next-generation intelligence for free through the Gemini app and AI Mode in Search. The era of fast, powerful, and accessible frontier AI is here.