Google has unveiled its new Gemini 3 Flash model, a faster and cheaper version of the Gemini 3 released last month. The company is making this model the default in the Gemini app and AI search mode, aiming to challenge OpenAI’s dominance. The release comes just six months after Gemini 2.5 Flash, but Google says the new version delivers far better performance and efficiency.
On benchmarks, Gemini 3 Flash showed strong results. It scored 33.7% on Humanity’s Last Exam, compared to 11% by Gemini 2.5 Flash and 34.5% by OpenAI’s GPT-5.2. On the multimodality and reasoning test MMMU-Pro, it outperformed all rivals with 81.2%. Google says the model is built to handle different types of content, from text and images to audio and video, making it more versatile for everyday use.
For consumers, Gemini 3 Flash is now the default model worldwide in the Gemini app. Users can still switch to the Pro model for advanced math and coding tasks. The new Flash can analyze videos, sketches, and audio recordings, and even generate visual answers with tables and images. It also supports app prototyping directly in the Gemini app. Meanwhile, Gemini 3 Pro is now available for search in the U.S., alongside the Nano Banana Pro image model.
On the enterprise side, companies like JetBrains, Figma, and Harvey are already using Gemini 3 Flash through Vertex AI and Gemini Enterprise. Developers can access it via API and Google’s new coding tool, Antigravity. Pricing is set at $0.50 per million input tokens and $3.00 per million output tokens, slightly higher than Gemini 2.5 Flash but faster and more efficient. The launch comes amid growing rivalry with OpenAI, which recently released GPT-5.2 after reports of declining ChatGPT traffic. Google says the competition is pushing all players to innovate and set new benchmarks in AI.