Tech Desk
May 14, 2024
Alphabet-owned Google introduced upgrades to its Gemini family of AI models at I/O, the company's annual developer conference held on Tuesday.
It's releasing a new model dubbed Gemini 1.5 Flash, which it claims is optimised for speed and efficiency.
"Gemini 1.5 Flash excels at summarisation, chat applications, image and video captioning, data extraction from long documents and tables, and more," said Google DeepMind CEO Demis Hassabis in a blog post.
Hassabis also stated that Google built Gemini 1.5 Flash because developers needed a model that was lighter and less expensive than the Pro version, which was introduced in February. Moreover, Gemini 1.5 Pro is more efficient and powerful than the company's initial Gemini model, which debuted late last year.
Gemini 1.5 Flash is positioned between Gemini 1.5 Pro and Gemini 1.5 Nano, Google's smallest variant that runs locally on smartphones.
Whereas, being lower in weight than the Gemini Pro, it is as strong. Google stated that this was accomplished through a "distillation" process in which the most important information and abilities from the Gemini 1.5 Pro were transferred to the smaller model.
In addition, this implies that Gemini 1.5 Flash will have the same multimodal capabilities as Pro, along with a one-million-token context window. According to Google, Gemini 1.5 Flash will be capable of simultaneously analysing a 1,500-page document or a codebase with over 30,000 lines.
The Gemini 1.5 Flash (or any of these variants) isn't intended for consumers. Instead, it's a faster and less expensive method for developers to create their own AI products and services with Google-designed technology.
Moreover, to introduce Gemini 1.5 Flash, Google is updating Gemini 1.5 Pro. The search engine giant stated that it has "enhanced" the model's ability to create code, reason, and analyse sounds and visuals.
However, the most significant improvement is yet to come: Google has revealed that the model's current context window will be doubled to two million tokens later this year. That would allow it to handle two hours of video, 22 hours of audio, over 60,000 lines of code, and over 1.4 million words at the same time.
Both Gemini 1.5 Flash and Pro are currently in public preview in Google's AI Studio and Vertex AI. The company also unveiled a new version of its Gemma open model, Gemma 2.
However, whether you are a developer or enjoy tinkering with constructing AI apps and services, these upgrades are not intended for the typical customer.
