Trending Articles
Trending Topics
Trending
Google is expanding its next-gen Gemini lineup with new “Flash” iterations of the large language model. And no, older readers, it has nothing to do with Macromedia.
The search giant announced that Gemini 2.0 Flash was released to general availability through the Gemini API in Google AI Studio and Vertex AI earlier this month.
Google describes Gemini 2.0 Flash as its “workhorse model for developers”, thanks to its lower latency and faster response times. The company is pitching Gemini 2.0 Flash as a model for “high-volume, high frequency tasks at scale”, with a context window of million tokens.
Google has also debuted Gemini 2.0 Flash-lite, a cheaper model, alongside the full-fat version.
When the company released an early version of Gemini 2.0 Flash in December last year, it reported that the experimental 2.0 Flash was outperforming 1.5 Flash on key benchmarks – including around coding, maths and reasoning – at twice the speed.
The latest version of Flash also packs new functionality compared to its predecessor, including multimodal outputs and inputs. Image generation and text-to-speech will be added to 2.0 Flash in the near future, Google said.
“Gemini 2.0 Flash’s native user interface action-capabilities, along with other improvements like multimodal reasoning, long context understanding, complex instruction following and planning, compositional function-calling, native tool use and improved latency, all work in concert to enable a new class of agentic experiences,” Demis Hassabisand Koray Kavukcuoglu, DeepMind CEO and CTO respectively, wrote on a recent blog.
Agentic AI is designed to be able to solve more complex problems and take actions on users’ behalf and Google is hoping that it will drive the next wave of AI uptake. The company said it is working on bringing agentic functionality to a number of projects currently under development, including Project Astra, its universal assistant, and a code agent for developers, Jules.
Consumers have been able to use the Flash version of Gemini 2.0 for a little while already: it began rolling out to both the mobile and desktop apps from the end of January.
Google has also added to the Flash family with 2.0 Pro Experimental, an early version of the Pro model which it says is better able to deal with complex prompts and understand and weigh up real-world knowledge. It sports a context window twice as large as non-Pro versions, at two million tokens.
