Google Updates Veo 3.1, Releases Gemini 2.5 Flash, Launches Presentation Creation In Canvas, And More

Google had long reigned as the undisputed titan of the web: master of search, information retrieval, mobile dominance, and early-stage AI.

Then the arrival of OpenAI’s ChatGPT hit like a thunderclap in Silicon Valley. Behind closed doors, Google’s leadership, and its army of engineers, felt a jolt they rarely, if ever, had before.

What unsettled Google wasn’t just the surge in public attention for ChatGPT, but the nature of the threat: a small-but-elegant chatbot whose underlying large-language-model architecture could converse almost like a human.

Google’s longtime predictability and algorithmic comfort suddenly felt exposed.

The internal atmosphere shifted. The company realized that the rulebook of the AI game had changed. Resources were redirected, focus was sharpened, and areas of weakness were acknowledged. Google’s crown was no longer assured.

Out of this pivot came ">Gemini. Google unveiled its next-generation model family, and the move signified it was no longer merely defending its turf, but actively rewriting the map.

And this time, Google has been busy with improving its AI, again.

First off, is an update to Veo 3.1.

If Veo 3 wasn't good enough, and Veo 3.1 didn't fulfill Google's ambitions, the update to the latter makes it even better.

According to Google, the update gives Veo 3.1 the ability to generate "new dimensions of detail." With "true to life textures, easier camera control, and dialogue with sound effects," Veo 3.1 can now create even more compelling stories.

Next, is the launch of Gemini 2.5 Flash.

This specific model is the "fast and cost-effective" member of the 2.5 family, delivering solid reasoning and multimodal capability, but optimized for volume and responsiveness. In comparison, Gemini 2.5 Pro is the "depth and precision" version, designed for bigger, more demanding tasks.

For users who have use-cases like frequent chat interactions, user support, or real-time assistance, Flash may hit the sweet spot. But if users are doing heavy-duty things, like advanced coding, detailed research or processing huge documents, Pro would be the go-to.

Them, complex formulas became easier than ever.

LaTeX rendering is now unified within a single workspace, allowing users to copy formulas straight into their documents, generate PDFs seamlessly, and edit them directly inside Canvas.

Presentations, too, received an upgrade.

Users could upload any source material to automatically generate full slide decks, complete with visuals and data visualizations, then export them to Google Slides for final adjustments. The feature started rolling out to Pro subscribers immediately, with Free users scheduled to gain access in the following weeks.

Entertainment also got smarter. Gemini on Google TV began curating personalized movie recommendations for every member of the family, while offering topic-based video suggestions tailored to each viewer’s interests.

Then, Google fixed an annoying inability, where users cannot switch AI models during an ongoing conversation without losing their chat history or starting from scratch.

With the update, users can finally change models mid-conversation, and use whatever Gemini models that fit their current query.

For example, the can use Flash for fast responses, Pro for everyday but heavy tasks, Gemini Ultra for more complex reasoning, or Gemini Nano for lightweight/on-device use.

To use this feature, the model selector allows users to switch model instantly. Future responses in that chat will use the new model, but past messages remain intact.

Google is moving fast, and it set of major announcements demonstrate its intention to reclaim and extend its lead in generative and multimodal AI.