Google is significantly expanding the capabilities of its Gemini AI, rolling out a new feature that allows the chatbot to analyze the content of user-uploaded videos. The multimodal update, which is becoming available to both free and paid users, lets you provide a video file and ask specific questions about its contents, from identifying objects to describing complex scenes, marking a major step in the AI’s ability to understand the world beyond just text and static images.
The feature is being deployed across Android, iOS, and the web, giving Gemini a notable advantage over competitors like ChatGPT, which currently lacks the ability to analyze uploaded video files. The rollout appears to be happening rapidly; after being spotted by 9to5google, more recent hands-on reports from Android Police confirm its availability on the web.
In one test, Gemini was able to accurately describe a scene from a forest and, using both visual and audio cues, correctly identify the location where the clip was filmed. While users can upload existing videos from their device’s gallery, the Gemini app’s built-in camera does not yet support direct video capture for analysis.
A Maturing and Tiered Model Family
This new capability is powered by Google’s Gemini 2.5 model family, which has recently shifted from a period of rapid experimentation to one of stability. The company’s Gemini 2.5 Pro and 2.5 Flash models are now in “General Availability,“ a status that, according to Vertex AI release notes, signifies they are stable and supported for production use.
This marks a strategic maturation from early 2025, when Google pushed its experimental 2.5 Pro model to all free users in what its social media team called a “sprint.”
That aggressive push drew criticism from some AI governance experts. The accompanying safety report arrived weeks later, which Kevin Bankston of the Center for Democracy and Technology called part of a “troubling story of a race to the bottom on AI safety.”
The new “GA” status suggests a more measured approach, underpinned by a newly defined hierarchy detailed on the Google for Developers Blog. In a post on The Keyword, Google Senior Director Tulsee Doshi framed the strategy: “to create a ‘family of hybrid reasoning models’ that deliver top-tier performance while remaining at the ‘Pareto Frontier of cost and speed.'”
The Gemini family now also includes the new Gemini 2.5 Flash-Lite, a cost-effective model in preview. It also features controllable “hybrid reasoning,” a system that allows developers to set a “thinking budget” for more complex queries, according to a post on the Google for Developers Blog.
A Simplified and Premium Subscription Strategy
The feature rollout is a key part of Google’s broader strategy to clarify its product lineup and drive adoption of its new premium subscription tiers. The company recently overhauled its confusing branding, consolidating everything under the simple “Gemini” name and introducing two distinct paid plans: Google AI Pro and Google AI Ultra. This move eliminated legacy branding like “Gemini Advanced” and “AI Premium.”
The Google AI Pro plan, at $19.99 per month, includes access to the powerful Gemini 2.5 Pro model, 2 TB of cloud storage, and monthly credits for other creative AI tools. The high-end Google AI Ultra plan, priced at $249.99 per month, is aimed at professionals and small studios, bundling 30 TB of storage, a YouTube Premium subscription, and exclusive access to experimental technology like the Project Mariner AI agent.
One executive described the Ultra tier as a “VIP pass” for users who want Google’s most advanced AI. The new structure, detailed on Google’s official plans page, also integrates and unlocks advanced features in other tools, such as the AI research assistant NotebookLM.
However, the value proposition is being debated. User feedback on Reddit suggests a key trade-off between the Pro plan and competitors like ChatGPT Plus, with Gemini favored by those deep in Google’s ecosystem and ChatGPT preferred for creative writing. Meanwhile, some analysts argue the high-end Ultra plan sets a new baseline for expensive AI subscriptions, as subscribers are essentially paying to beta test features.
The Expanding Creative Toolbox and Competitive Landscape
While video analysis is the latest addition, it joins a growing suite of creative tools that Google is using to justify its subscription costs. The AI Pro and Ultra plans also grant access to the company’s advanced text-to-video generation model, Veo 3. This technology, which Google DeepMind CEO Demis Hassabis memorably described as helping the industry “emerge from the silent era of video generation,” is accessible through the specialized Flow filmmaking app and is also being integrated into Google Vids.
Veo’s main advantage is its native, synchronized audio generation, while Sora excels at longer clip generation and physics simulation. To address concerns about authenticity and misuse, Google is using its SynthID technology to apply an imperceptible digital watermark to AI-generated content.
These safety measures are critical, as Veo 3 can generate convincing deepfakes of sensitive events like riots and election fraud. This technological arms race is a central theme of the AI era, unfolding as part of the massive, ecosystem-wide integration strategy on full display at Google’s I/O 2025 conference.
Furthermore, questions about training data persist, with reports that models like Veo may be trained on public YouTube content. This contrasts with competitors like Adobe, which emphasizes that its Firefly models are trained on licensed content to avoid intellectual property issues.
