On November 18 (local time), Google formally introduced its newest artificial intelligence model, Gemini 3. On its launch day, this model was seamlessly integrated into Google Search, the Gemini app, and numerous developer platforms. Moreover, it was deployed across a range of profitable products. Sundar Pichai, CEO of Google's parent company, Alphabet, emphasized that the new model is designed to deliver more accurate and comprehensive answers to complex queries. This means users can achieve their desired outcomes with fewer input prompts.

In addition, Gemini 3 has witnessed substantial improvements in programming, app development, and image - generation capabilities. Starting from its release date, the model will be accessible to a select group of subscription users, with plans to gradually expand its availability in the upcoming weeks.

When it comes to performance, Gemini 3 has made a remarkable mark. It secured the top position on the global AI model LMArena leaderboard, scoring 1501. In the Humanity's Last Exam benchmark test, which evaluates general reasoning abilities, it achieved an impressive score of 37.5%, outperforming the previous record set by GPT - 5 Pro. The model has also established new industry benchmarks in multiple areas, such as mathematics, multimodal understanding, and factual accuracy.

Google also unveiled the Gemini 3 Deep Think enhanced reasoning mode. This mode has excelled in various tests, showcasing its capability to handle unfamiliar challenges. Furthermore, Google introduced a brand - new agent - first development platform, Google Antigravity. By leveraging Gemini 3's advanced reasoning, tool - utilization, and agent - coding capabilities, this platform aims to transform AI assistance. Instead of being just a tool in developers' kits, AI will now become an active and collaborative partner.