On December 29, 2025, Google published a recap of its December AI announcements on its official blog. The update spans Gemini models, verification tools, translation features, search expansions, and media recaps. These changes affect the Gemini app, Google Search, Google Translate, YouTube, Google Photos, and Shopping in multiple markets. A full video version of the recap is also available for those who want to download it.
Key Details
- Gemini 3 Flash rollout: Google introduced Gemini 3 Flash, describing it as “frontier intelligence built for speed” for everyday tasks. It is rolling out as the default model in the Gemini app and in AI Mode in Search for users worldwide.
- Video verification in the Gemini app: Google added video verification tools that use SynthID watermarks in the Gemini app. People can upload videos up to 100 MB or 90 seconds and ask whether the content was generated or edited using Google AI.
- Browsing and research tools: Google Labs announced Disco, a browsing experiment focused on complex online tasks. Disco features GenTabs, which synthesizes open tabs and Gemini chats into interactive web applications. Google also released a more powerful Gemini Deep Research agent through the Interactions API. To support evaluation, Google open-sourced DeepSearchQA for testing research agents on web tasks.
- Gemini audio models and live translation: Google upgraded its Gemini audio models, including Gemini 2.5 Flash Native Audio, for more natural voice interactions. The model is available in AI Studio, Vertex AI, Gemini Live, and for the first time in Search Live. A live speech translation beta in the Google Translate app delivers real-time translation in more than 70 languages, sending audio directly to users’ headphones while preserving original intonation and pacing.
- Virtual try-on updates for U.S. shoppers: Google introduced a new way for U.S. shoppers to use its virtual try-on tool. Instead of a full-body photo, users can upload a selfie, and Nano Banana generates a realistic full-body digital version. After selecting a studio-like image and clothing size, shoppers can see that digital version wearing products from the Shopping Graph.
- Gemini 3 and Nano Banana Pro in Search: Google brought its Gemini 3 model to AI Mode in Google Search across nearly 120 countries and territories in English. Google AI Pro and Ultra subscribers can access Gemini 3 Pro via “Thinking with 3 Pro” in AI Mode. The Nano Banana Pro generative imagery model is also available in AI Mode in more countries in English.
- Year-end recaps across YouTube, Photos, and Search: YouTube’s 2025 trends summary named MrBeast top creator for the sixth straight year. It also reported Rosé and Bruno Mars’ track “APT.” as the fastest K-pop video to hit one billion views. YouTube introduced a personalized Recap, Google Photos Recap gained new controls, and Google released the Year in Search 2025 report.
Background Context
Google states it has spent more than 20 years investing in machine learning and AI research, tools, and infrastructure. According to the company, teams apply AI in areas such as healthcare, crisis response, and education worldwide. The December recap continues a regular series of posts summarizing recent Google AI product updates and experiments.
Source Citations
- December recap and Gemini 3 Flash: December AI recap, Gemini 3 Flash announcement.
- Gemini app and AI Mode updates: Gemini 3 Flash in Gemini app, AI Mode update, video verification tools.
- Browsing and research tools: GenTabs and Disco, Deep Research agent, DeepSearchQA and Gemma uses.
- Audio and translation updates: Gemini audio models, Search Live audio, Translate speech features.
- Shopping and Search models: Virtual try-on tool, Gemini 3 and Nano Banana Pro in Search.
- Media and trends recaps: YouTube 2025 trends, YouTube Recap, Google Photos Recap, Year in Search 2025.
.svg)
.svg)
.svg)