Google Releases Gemini 3 Flash: Frontier AI at Flash Speed

Google announced Gemini 3 Flash on December 17, 2025, expanding its Gemini 3 model family with a model that delivers frontier-level intelligence at three times the speed of its predecessor. The new model is now the default across the Gemini app and AI Mode in Search, bringing advanced AI capabilities to millions of users globally at no cost. Gemini 3 Flash achieves 90.4% on GPQA Diamond (PhD-level reasoning) and 78% on SWE-bench Verified (coding), outperforming even Gemini 2.5 Pro while using 30% fewer tokens on average. The model's pricing structure reflects Google's push for accessibility: $0.50 per million input tokens and $3.00 per million output tokens. This positions it at the cost frontier while maintaining performance that rivals larger, more expensive models. Developers can access Gemini 3 Flash through the Gemini API, Google AI Studio, Vertex AI, and the newly launched Google Antigravity platform, with a free tier available for both UI and API access.

Google's API infrastructure is now processing over 1 trillion tokens daily since the Gemini 3 family launch, demonstrating substantial developer adoption. Early enterprise adopters including JetBrains, Bridgewater Associates, and Figma are already integrating the model into production workflows. The speed advantage proves particularly valuable for agentic coding tasks, real-time multimodal analysis, and interactive applications requiring both quick responses and sophisticated reasoning.

The model's multimodal capabilities enable real-time video analysis, audio processing for educational assessments, and visual understanding while users sketch or draw. In the Gemini app, users can upload video content and receive actionable plans within seconds, or describe an app idea by voice and watch Gemini 3 Flash build a working prototype in minutes. These applications highlight how the model balances computational efficiency with the advanced reasoning previously reserved for larger, slower models.

For Search integration, AI Mode powered by Gemini 3 Flash can handle complex queries with multiple considerations, from last-minute trip planning to educational concept breakdowns. The model parses nuanced questions and delivers comprehensive, visually digestible responses by pulling real-time information from across the web. This effectively combines research depth with the immediacy users expect from Search, without sacrificing the reasoning quality introduced with Gemini 3 Pro.