Not to be outdone by OpenAI's Sora drop, Google just released its new AI model Gemini 2.0.
On Wednesday, Google introduced Gemini 2.0 Flash, the first family member of the next generation of AI models. Gemini 2.0 Flash is described in the announcement as a "workhorse model" for developers, capable of powerful performance at scale. Flash supports image and audio generation, has native integration with Google Search, can write code, and works with third-party apps. Alongside the Gemini 2.0 Flash announcement, Google also introduced Deep Research, a Gemini feature that browses the web and compiles research reports based on the initial prompt.
Gemini 2.0 Flash is a step up from Gemini 1.0 in that has improved reasoning, longer context windows, ability to understand complex instructions, and native tool use — all of which has been designed to make the model more agentic, in other words capable of executing multi-step tasks on the user's behalf.
As part of this, Google said Gemini 2.0 would be available for Project Astra, a research prototype for testing a universal AI assistant. Google also shared other research prototypes: Project Mariner, which is specifically designed to explore "human-agent interaction," and Project Jules for developers.
Gemini 2.0 Flash is available as an "experimental model" via the Gemini API which can be accessed in Google AI Studio and Vertex AI. But casual users can also try out its improved chat capabilities in the Gemini desktop app, with mobile app support coming soon.
