Google is not stopping, it has hit the accelerator with its generative AI: After Google killed Bard and introduced the Gemini Advanced paid subscription, all based on version 1.0 of its model, The company already has Gemini 1.5 ready. Faster, more powerful, and in a change that could make a real difference: Google has greatly expanded the context in which its AI can operate.

It can’t be said that Google, a pioneer in training, analysis and execution tools, has done little for artificial intelligence. But it was clear Google could not offer this artificial intelligence to the consumerOpen AI has taken over the baton of productive chatbots with the ubiquitous ChatGPT. And of course, Google couldn’t be left behind, so it stepped up with Gemini. So much so that the new version of the multimodal model is already ready.

Gemini 1.5: up to 1 million coins compared to 128,000 coins for GPT-4

Token comparison. google official

The numbers that summarize how upset Google is with the dominance of Open AI and GPT models are numbers that reveal Google’s impressive theoretical potential. new Gemini model, recently released version 1.5. Google guarantees that it achieves very high results in AI benchmarking; It provides an 87% advantage over the previous version of the model, Gemini Pro 1.0.

Gemini 1.5 can consistently handle media containing up to one million tokens. This takes the understanding of context and model response far beyond existing generative AI tools. To contextualize, GPT-4 Turbo supports up to 128,000 coins. And Gemini Pro 1.0, the version Google currently uses in its free public AI, uses 32,000 coins.

With the new multi-modal model, the user will have better responses to commands, greater creativity in behavior, Gemini 1.5 will understand the context better and will also be faster in execution; fewer resources are needed for production tasks.

As Google explains on its blog:

“Gemini 1.5 Pro can process large amounts of information at once, including 1 hour of video, 11 hours of audio, code bases containing more than 30,000 lines of code, or more than 700,000 words. In our research, we have successfully tested up to 10 million tokens.”

The new generative AI model will not be publicly available yet, but developers and some enterprise customers will have access to it soon. Tests will be carried out on Gemini Pro 1.5 with a limit of 128,000 tokensthe same ones currently handled by Gemini Ultra 1.0.

Once the testing period is over, Google will move its multimodal language model into three parts: Nano (on mobile phones like the Pixel 8 Pro), Pro (on the free version of the Gemini chatbot), and Ultra (Gemini Advanced). No date has been given for the update to all tools currently available for individual and commercial users.

Xataka on Android | The Gemini app for Android is here. So you can try Google’s most advanced artificial intelligence on your mobile phone