Google's Gemini 2.0 Now Available; Flash To Pro, All You Need To Know

Google has made Gemini 2.0 available in Google AI Studio and Vertex AI, and in the Gemini app for Gemini Advanced users. The company actually confirmed via its blog the availability of as many as three experimental models. This includes Flash, Flash-Lite and Pro. With DeepSeek ramping up competition, Google is looking to grab the initiative even as the nimble Chinese rival put the spotlight on itself due to its amazing performance metrics developed at surprisingly low cost.

Google revealed the highlights as it expanded Gemini 2.0 for developers.

First is Gemini 2.0 Flash, which Google claimed comes with higher rate limits, stronger performance, and simplified pricing. The information technology firm said that developers can now build production applications with Gemini 2.0 Flash.

Then there is the Gemini 2.0 Flash-Lite, which Google has dubbed as being its most cost-efficient model yet. It indicated that Flash-Lite is now available in public preview.

And last, but not least, is the Gemini 2.0 Pro model. Google indicated that this is an "experimental update to our best model yet for coding and complex prompts". Pro is now now available, it added.

Google indicated that Flash models are adept at handling many tasks quickly and efficiently, especially when dealing with large quantities of information. Flash models can understand different types of data such as text and images and are able to handle up to 1 million tokens of information at once.

For those who are interested, Google had some extra goodies to offer. The tech major said that along with the Gemini 2.0 Flash Thinking Experimental, which was rolled out recently, "...the Flash variant that reasons before answering, these releases make Gemini 2.0 capabilities available to a broad range of use cases and applications."

"All of these models will feature multimodal input with text output on release, with more modalities ready for general availability in the coming months," the company said.

The firm's Gemini 2.0 lineup was reportedly built with new reinforcement learning techniques that use Gemini itself to critique its responses.

The firm said that this process leads to more accurate and targeted feedback and improved the model's ability to handle sensitive prompts.

Google reported using 'automated red teaming', which is a technique where a software is used to simulate cyberattacks on the application, to find weak spots in its security.

The firm also put in safety measures to protect against indirect prompt injection, a type of cybersecurity attack which involves attackers hiding malicious instructions in data that is likely to be retrieved by an AI system.

Google will also be releasing an experimental version of Gemini 2.0 Pro, calling it their "best model yet for coding performance and complex prompts."

And then there is pricing. Google is saying that it focused on slashing costs with Gemini 2.0 Flash and 2.0 Flash-Lite. In fact, it indicated it was breaking from the previous model. It indicated that 2.0 Flash and Flash-Lite price can be lower than Gemini 1.5 Flash and this is despite the improved performances.