Google’s Big Gemini AI Updates: AI Models, Search, Preferred Sources and More From the Week

Google announced several upgrades to its Gemini text-to-speech models alongside new tools and commercial partnerships with news publishers on Thursday. The biggest highlight from the announcements is an upgrade to Gemini 2.5 text-to-speech artificial intelligence (AI) models that gives developers nuanced control over voice output, including tone, expressiveness and multi-speaker consistency. The tech giant also detailed improvements to AI Mode and Search with more inline links to sources and personalised news features. Preferred sources, which was introduced earlier this year, is now being rolled out globally.

Google Brings New AI Features

In a blog post, the Mountain View-based tech giant announced a new Gemini 2.5-based text-to-speech models. Available as a preview, Gemini 2.5 Flash TTS and Gemini 2.5 Pro TTS are now available to developers. The updated models now have improved vocal expressions that sound natural and can be adjusted to generate a range of stylistically different audio output.

Developers can now guide the model to match specific tones and speaking styles, adjust pacing dynamically, and manage transitions between distinct character voices in multi-speaker content. Google says the models can now be used for more complex use cases, such as interactive virtual assistants, audiobooks, podcasts and localised audio narratives.

Multi-speaker handling has also been strengthened, offering more consistent voice identity across interviews or scripted dialogue. Developers and content creators can experiment with accent and style variations to suit regional preferences or narrative requirements, the tech giant claimed. These AI models are now available via the Gemini application programming interface (API) in Google AI Studio.

Google also outlined a series of developments designed to make Search features more responsive to user preferences and supportive of the web content ecosystem. Among the changes is the global rollout of a “Preferred Sources” option in Search, which allows users to customise news and content feeds so that results from selected websites appear more frequently.

The company also plans to make news subscription links more visible in its AI-powered experiences across the Gemini app and AI Mode in Search. So, when users see the responses, they will see more sources from their subscribed publishers. The design has also been changed with dedicated highlights and contextual placements, similar to news articles. The overall number of inline links within the responses have also been increased.

To support these features, Google has expanded the Web Guide experiment so it appears more frequently in search results and performs faster, organising links into thematic groups to help users discover related information.

Additionally, Google said it is piloting commercial partnership programmes with a range of media organisations, including major global and regional publishers, to explore how AI tools can support audience engagement on news platforms. The initiative encompasses AI-powered article overviews and audio briefings with clear attribution and hyperlinks to original content. Google has also partnered with real-time information providers to improve query results in the Gemini app.