Google Gemini is a family of advanced multimodal artificial intelligence (AI) models developed by Google. Introduced as the successor to Google Bard, Gemini is designed to be an AI assistant capable of understanding and processing various types of input, such as text, images, audio, video, and programming code. Since its official launch in February 2024, Gemini has evolved into one of the most comprehensive AI ecosystems, deeply integrated with Google services and available in various subscription tiers to meet the needs of users ranging from students to enterprises.
History and Development
The roots of Google Gemini lie in Google Bard, a chatbot based on the LaMDA language model that was launched in 2023. A major transformation took place in early 2024 when Bard was upgraded with a more advanced next-generation large language model (LLM). On February 8, 2024, Google officially announced the name change from Bard to Gemini, marking a strategic shift to unify all of its AI capabilities under a single brand umbrella. This change was also accompanied by the launch of the Gemini app for smartphones, which gradually replaced Google Assistant as the primary digital assistant on certain Android devices.
Throughout 2025 and 2026, Google continued to develop Gemini models at a high pace. The launch of models such as Gemini 3.5 Flash and Gemini 3.1 Flash-Lite demonstrated Google's commitment to providing models with a balance between high intelligence and low latency, especially for programming and agentic tasks. Recent innovations include the introduction of Gemini Omni Flash, a multimodal model capable of generating 720p-quality video from text descriptions or editing video through simple conversation. This model was released in public preview in late June 2026, marking a major step for Google in the field of generative content creation.
Core Capabilities and Features
Multimodality and Content Processing
Gemini's main feature is its capability as a multimodal model. Users can interact with Gemini through text, voice, and images. The Gemini app allows users to take photos and ask questions about objects around them. This capability is extended with the video-to-image feature on the latest model, in which Gemini can generate high-quality thumbnails or posters from uploaded video files. Gemini is also able to create Audio Overviews, turning documents and reports into engaging podcast-style discussions, making complex content easier to consume.
Integration and Extensions
Gemini is designed to work seamlessly with the Google ecosystem. Through the extensions feature, Gemini can connect with apps such as Gmail, Google Maps, YouTube, and Google Drive to help complete tasks more quickly. This capability allows Gemini, for example, to summarize email contents, search for travel routes, or find information from videos without having to open the apps separately. On Pixel 9 devices and later, Gemini is the default assistant, accessible by long-pressing the power button or using the voice command "Hey Google".
Personalization with Gems
One of the main personalization features is Gems. Gems are specialized versions of Gemini that can be customized to become "experts" on a particular topic without requiring coding skills. Users can create Gems for specific purposes such as writing editors, event planners, or sentiment analysis, by uploading instructions and reference files. Google also provides ready-made Gems, such as Learning coach for study guidance and Career guide for career preparation, which are highly useful in the education sector.
Gemini Live and Conversation
The Gemini Live feature delivers a more natural conversational experience. Users can speak with Gemini in real time, use it to brainstorm ideas, or simulate important conversations. Gemini Live is designed for smooth interaction, allowing users to share their screen or camera while speaking in order to receive contextual responses.
Models and Versions
Google releases Gemini in various model variants to meet different needs and budgets. Each model has optimized specifications, ranging from speed and cost to intelligence and multimodal capability.
Gemini 3.5 Flash
This model is a mainstay for tasks that require substantial intelligence at high speed. Released generally in May 2026, Gemini 3.5 Flash is known to be highly effective for programming tasks and long-term "agentic" tasks, offering performance comparable to large "flagship" models at a lower cost.
Gemini 3.1 Flash-Lite
This variant is optimized for speed, scale, and cost efficiency. The model is well suited to simple tasks and image generation with ultra-low latency.
Gemini Omni Flash
The newest model, still in public preview. Its main capability is conversational video creation and editing. Equipped with an invisible SynthID digital watermark for content verification, this model represents a new generation of generative AI capable of creating content from various types of input.
Image Models (Nano Banana)
Gemini also has specialized visual models such as Gemini 3.1 Flash Image ("Nano Banana 2") and Gemini 3 Pro Image ("Nano Banana Pro"). These models have been generally available and excel at image creation and editing, including a feature for generating images from video.
Availability and Subscriptions
Gemini can be accessed through various platforms, including mobile apps for Android (Android 10+) and iOS, as well as through the website at gemini.google.com. Google offers several subscription plans.
The Google AI Pro plan unlocks access to the most advanced models such as 2.5 Pro, the Deep Research feature for detailed reports, and eight-second video generation with Veo 3, as well as a 1-million-token context window capable of processing up to 1,500 pages of text.
The Google AI Ultra plan is the highest tier, offering exclusive access to the most powerful models, the latest features such as Agent Mode, and the opportunity to try Google's AI innovations earlier. This plan is available in certain countries and is not for Workspace customers.
In Southeast Asia, Gemini has recorded significant user growth, more than doubling in a year. Vietnam has become the regional leader in using Gemini for academic purposes and local-language use. The country has even become the market with the highest percentage of native-language (Vietnamese) use in the region, reaching 89%.
Impact and Education
Gemini, along with other generative AI, has begun to change the way people access information. This change has sparked discussion about the future of traditional knowledge sources such as Wikipedia. Wikimedia Indonesia, for example, held the #WikipediaxAI program in 2025 to discuss how AI uses Wikipedia content and how the community can adapt. Concerns have arisen that the shift from reading articles to receiving instant summaries from AI could reduce the number of visitors to source sites and the number of new volunteer contributors.
On the other hand, Gemini offers great potential in education. In Vietnam, more than 160,000 students use the Gemini Canvas feature every month for exam preparation, while educators generate tens of thousands of teaching assistance requests every day. This use shows AI's role as a powerful supporting tool, not a replacement for human critical thinking. A comparative study even placed Gemini as a tool comparable to ChatGPT for various translation tasks, demonstrating its capability in language processing.
References
1. Use Gemini, your AI assistant, on your Pixel phone. Google Help.
2. Education/News/December 2025/WikipediaxAI: Wikipedia, AI, and the future of knowledge. Meta-Wiki, Wikimedia.
3. Catatan rilis, Gemini API. Google AI for Developers.
4. Personalize Gemini with Gems: custom AI experts for any topic. Google Services.
5. Google Gemini, Apps on Google Play.
6. Cara Menggunakan Gemini AI Google Bahasa Indonesia, Mudah dan Praktis. Kompas.com.
7. Release notes, Gemini API. Google AI for Developers.
8. Gemini, "Asisten" Canggih untuk Sehari-hari. LinkedIn, 360 Solusi Teknologi.
9. Việt Nam dẫn đầu Đông Nam Á về dùng AI của Google hỗ trợ học thuật. VietnamPlus.
10. Digilib UIN SGD (Bab 1).
11. 100 things we announced at I/O 2026. Google Blog.
12. Try Gemini, your personal AI assistant. Android.com.