Google's Gemini AI assistant continues to evolve rapidly with significant new capabilities being discovered and rolled out. Recent findings reveal upcoming video analysis capabilities and chat history search functionality, while the latest Gemini 2.5 Pro model has impressed users enough to join the ranks of paid AI tools worth subscribing to.
Video Analysis Coming to Gemini
An APK teardown of Gemini for Android (version 16.15.38.sa.arm64 beta) has revealed that Google is preparing to add video analysis capabilities to its AI assistant. Developers were able to enable this unreleased feature, allowing users to upload videos and ask Gemini questions about their content. The implementation appears to handle up to five minutes of video content, with the AI accurately analyzing visual elements, identifying locations based on architectural cues, and providing contextual information about the footage.
This functionality could prove useful for summarizing recorded meetings, troubleshooting technical issues via screen recordings, or analyzing visual content for various purposes. The feature appears to be in final testing stages, suggesting an official announcement could come soon, possibly at Google's upcoming I/O event.
Upcoming Gemini Features
- Video analysis with 5-minute upload limit
- Chat history search functionality
- Improved document analysis capabilities
Search Functionality for Chat History
Another notable improvement discovered in the same beta version addresses a significant usability gap in Gemini. Currently, while the AI maintains a comprehensive history of past conversations, there's no efficient way to search through them. This omission has been particularly frustrating for heavy users trying to locate specific information from previous interactions.
The APK teardown revealed strings related to SEARCH_CHAT_HISTORY, indicating Google is developing a solution to this problem. While the feature appears to be in early development stages with no functional implementation yet visible, it represents an important quality-of-life improvement that aligns with Google's core competency in search technology.
Gemini 2.5 Pro Impresses Users
Beyond these upcoming features, Google's recent release of Gemini 2.5 Pro has been making waves among AI enthusiasts. The model represents a significant advancement over previous versions, with capabilities that include complex reasoning, coding, and analyzing large documents. Its massive context window of up to 1 million tokens (with plans to expand to 2 million) allows it to process and understand much more information at once than earlier versions.
Alongside Gemini 2.5 Pro, Google also introduced Gemini 2.5 Flash, optimized for speed and cost efficiency in high-volume applications like customer service. Both models support multimodal input across text, images, audio, and video, with features like Manual Reasoning Control for adjusting how thoroughly the model thinks through problems.
The improvements have been substantial enough that even selective AI users who typically limit their paid subscriptions are adding Gemini to their toolkits. The key to getting the most from these advanced models appears to be in how users approach them—with clear communication, specific prompts, and realistic expectations about their capabilities and limitations.
Gemini 2.5 Pro Features
- Context window of up to 1 million tokens (2 million coming soon)
- Multimodal input support (text, images, audio, video)
- Manual Reasoning Control for adjusting model thinking
- Canvas: interactive space for document and code editing
- Live API for real-time streaming across modalities
The Evolving AI Landscape
While Gemini 2.5 Pro represents a major step forward for Google's AI offerings, users note that the integration within Google's ecosystem still feels somewhat disjointed when switching between different functions. The model also continues to face challenges with nuanced writing, deeper thinking, and occasional hallucinations—issues common across large language models.
As AI tools continue to mature, the distinction between them becomes less about raw capabilities and more about how effectively users can harness them for specific workflows. With these upcoming features and recent improvements, Google is positioning Gemini as a serious contender in the increasingly competitive AI assistant space, particularly for users who value multimodal analysis capabilities and deep integration with Google's ecosystem.