Google has launched the Gemini 1.5 Pro AI model, outperforming the competition

Google DeepMind

Google has unveiled a new model of artificial intelligence Gemini 1.5 Pro, which, according to technical experts, shows better results than competing developments in a number of benchmark tests.

Here's What We Know

Gemini 1.5 Pro belongs to the class of multimodal models capable of handling text, images, audio and video. According to data published by Google, the new model performs almost perfectly in information retrieval tasks on long texts of up to 10 million tokens.

In comparison, existing models such as Claude 2.1 and GPT-4 Turbo can only handle contexts of 200,000 and 128,000 tokens, respectively.

In addition, Gemini 1.5 Pro has demonstrated outstanding capabilities in long text analysis, long duration video processing, speech recognition and other areas. For example, the model is able to analyse a 45-minute film and answer a specific query about its content.

For now, Gemini 1.5 Pro is available to developers in test mode via Google's cloud services. The company promises to soon introduce a model for widespread use with a context window limit of 128,000 tokens.

Source: Google DeepMind