Google Releases Its Most Powerful AI Model, Gemini 3—Here's What You Need to Know

Google has launched Gemini 3 Pro, its most advanced AI model yet, capable of processing text, images, audio, and video simultaneously with a context capacity of up to 1 million tokens. In public tests, Gemini 3 Pro surpassed its predecessor, Gemini 2.5 Pro, achieving scores of 37.5% on academic reasoning tasks versus 21.6% from the earlier version. This model employs a sparse mixture-of-experts architecture, activating only specialized subnetworks for given tasks, thus optimizing performance and computational efficiency. The human-like interactive capabilities of Gemini 3 focus on coding tasks, offering detailed suggestions to users, aiding in error fixing, and enhancing app performance. However, the model does have limitations, such as potential hallucinations and speed issues, and it is currently not available on the consumer-facing Gemini app. The public preview is open and free for everyone, while Google aims to compete against leading models like OpenAI's ChatGPT, which boasts significant user numbers.

Source 🔗