3 Minutes
Google has unveiled Gemini 3, the newest generation of its AI models designed to be clearer, more candid, and better at understanding complex inputs. The rollout begins with preview releases for paid users and broader availability in the Gemini app.
A smarter assistant that tells you what you need to hear
Gemini 3 focuses on delivering concise, insightful responses rather than flattering or vague replies. Google says the model trades cliched praise for substantive answers, aiming to give users direct guidance and real context with less prompting. Imagine asking for career advice and getting a focused action plan instead of empty encouragement — that is the promise here.
Multimodal power: text, images, video, audio and code
Built from the ground up for multimodal understanding, Gemini 3 can synthesize information across text, images, video, audio, and even code. That means richer visualizations, deeper interactivity, and the ability to pull together insights from multiple content types in a single response.
- Richer visual outputs for images and video
- Cross-modal synthesis — combine text with images or code snippets
- Faster, more relevant results with fewer follow-up prompts
Rollout, tiers and the Deep Think variant
Gemini 3 is launching first in AI Mode within Google Search for Google AI Pro and AI Ultra subscribers, while the Gemini app will get the new model for all users. The initial release is Gemini 3 Pro, currently in preview. A more powerful edition, Gemini 3 Deep Think, is under safety testing and will be offered to AI Ultra subscribers once cleared.
According to Google, Gemini 3 Pro outperforms Gemini 2.5 Pro across major benchmarks, and Deep Think beats Pro on tests too — at the cost of longer response times when more deliberation is needed.
Safety, benchmarks and real-world use
Google describes Gemini 3 as its most secure model so far, saying it has gone through the most comprehensive safety evaluations to date. The company highlights reduced sycophancy, stronger resistance to prompt-injection attacks, and improved safeguards against misuse. While those claims will be scrutinized by the AI community, the emphasis on safety reflects growing expectations for responsible AI deployment.
For users and developers, Gemini 3 promises a noticeable step forward in quality: smarter answers, broader multimodal skills, and tiered access for consumers and professional subscribers. Whether you use it in Search, the Gemini app, or via Google subscription tiers, expect interactions that aim to be more useful and less performative.
Source: gsmarena
Leave a Comment