Google has given access to an early version of its latest AI model Gemini to a small group of companies, The Information reported (paywall) on Thursday, citing three unnamed sources with the direct knowledge of the matter.
This follows another report by The Information last month, which claimed that the company is expected to launch the new model this fall.
“Giving outside developers access to the software, known as Gemini, means Google is getting close to incorporating it in its consumer services and selling it to businesses through the company’s cloud unit,” noted the latest report.
The model is expected to compete with GPT-4, OpenAI’s most powerful model to date. The version Google has provided access to is a fairly large version of Gemini, but it is not the largest, noted the report.
Gemini is a group of LLMs which is expected to power company’s AI chatbot Bard which currently runs on Palm 2, a large language model Google had launched earlier this year. The upcoming model is also expected to be used in Google’s other products, including its enterprise apps, like Google Docs and Slides.
The previous reported stated that Gemini will be able to summarize or generate text, create contextual images, and eventually perform complicated tasks like analyzing a flow chart or controlling a software with their voice. The model would also help software engineers write code.
Sundar Pichai, the CEO of Google, in May had stated the the model was in training, “Gemini was created from the ground up to be multimodal, highly efficient at tool and API integrations and built to enable future innovations, like memory and planning. While still early, we’re already seeing impressive multimodal capabilities not seen in prior models. Once fine-tuned and rigorously tested for safety, Gemini will be available at various sizes and capabilities.”