Jakarta, domclub Indonesia
—
Google
released their latest artificial intelligence (AI) model,
Gemini 3
, which is claimed to be the world’s best model for multimodal understanding.
“Today we took another big step on the path to AGI and released Gemini 3,” said Koray Kavukcuoglu, CTO of Google DeepMind and Google’s Chief AI Architect in an online preview session, Tuesday (18/11).
ADVERTISEMENT
SCROLL TO CONTINUE WITH CONTENT
Koray claims Gemini 3 is the world’s best model for multimodal understanding.Apart from that, Gemini 3 is also called the best model for Google agentic and vibe coding.
He said this AI model provides richer visualization and deeper interactivity.
The Gemini 3 era began with the launch of Gemini 3 Pro in preview and is available today across a variety of Google products, from Search to the Gemini chatbot.
Google also introduced Gemini 3 Deep Think, an improved reasoning mode to push Gemini 3’s performance even further.Google provided access to security testers before making this model available on the Google AI Ultra subscription option.
Google CEO Sundar Pichai said Gemini 3 is a technology at the forefront of reasoning, and is designed to understand depth and nuance, whether it’s capturing subtle clues in creative ideas, or unraveling overlapping layers in complex problems.
He claims Gemini 3 is far superior at understanding the context and intent behind user requests, so users get what they need with just a few hints.
“Like previous generations, Gemini 3 is once again advancing the boundaries of the latest technology. In this new chapter, we will continue to push the boundaries of artificial intelligence, agents and personalization so that AI truly benefits everyone,” he said in a statement.
Ability
Gemini 3 Pro is said to be able to realize any idea with its reasoning and multimodal capabilities.Its performance is far superior to the 2.5 Pro on every major AI benchmark.
Google’s internal testing claims the Gemini 3 Pro tops the LMArena leaderboard with a breakthrough score of 1501 Elo.The model also demonstrated PhD-level reasoning abilities with the highest scores on Humanity’s Last Exam (37.5 percent without using any tools) and GPQA Diamond (91.9 percent).
Gemini 3 is also said to set a new standard for leading models in mathematics, achieving a new leading score of 23.4% on MathArena Apex.
Beyond text, Gemini 3 Pro redefined multimodal reasoning with scores of 81 percent on MMMU-Pro and 87.6 percent on Video-MMMU.The model also achieved a leading score of 72.1% on SimpleQA Verified, demonstrating significant progress in factual accuracy.
Koray said this data shows the Gemini 3 Pro is very capable of solving complex problems in various topics such as science and mathematics with a high level of reliability.
In Search, AI Mode now uses Gemini 3 to deliver new generative user interface experiences, such as immersive visual layouts and interactive tools and simulations.
For developers, Gemini 3 is said to be here to boost productivity.Gemini 3 is said to have topped the WebDev Arena leaderboard with an impressive score of 1487 Elo.
The Gemini 3 also scored 54.2 percent on Terminal-Bench 2.0, which tests the model’s ability to use tools to operate a computer via a terminal, and far outperformed the 2.5 Pro on SWE-bench Verified (76.2 percent), a benchmark that measures programming agents.
Developers can now develop with Gemini 3 on Google AI Studio, Vertex AI, Gemini CLI, and the new agent development platform, Google Antigravity.This model is also available on third-party platforms such as Cursor, GitHub, JetBrains, Manus, Replit and others.
(lom/dmi)
[Gambas:domclub Video]
