Gemini 2.0 Flash is Google's latest response to OpenAI's o3 and DeepSeek's R1 model
With the upswing in AI advancements following DeepSeek's prominent entry into the artificial intelligence arena in January, Google like many other competitors in the AI landscape, is accelerating its own AI initiatives, launching a series of new AI models under the Gemini umbrella.
At the forefront, is Gemini 2.0 Flash, a direct response to OpenAI's o3 and DeepSeek's R1 models. If you remember, this model dropped late last year as an experimental release. Since its initial release, it has undergone significant updates. We’re talking multimodal output, imagine images mixed with text and even multilingual audio, say steerable text-to-speech (TTS).
The model now also has improved reasoning capabilities, and can natively call tools like Google Search, code execution, and third-party user-defined functions. It also has reduced latency and is now generally available via the Gemini API in Google AI Studio and Vertex AI.
Google did not stop there. It also rolled out an experimental version of Gemini 2.0 Pro, which is anticipated to be its most advanced model yet. The 2.0 pro model is specifically tailored for coding and handling complex prompts. Imagine processing a massive 2 million token context window. If you’re a Gemini Advanced subscriber, you can already play around with it via the model-picker drop-down menu on both desktop and mobile. Developers can also access it in Google AI Studio and Vertex AI.
If you don’t want to break the bank, well, Google also introduced Gemini 2.0 Flash-Lite specifically as a cost-efficient model. Despite being cheap, it doesn’t skimp on performance, outperforming its predecessor, 1.5 Flash, on several benchmarks, with a 1 million token context window and multimodal input capabilities.
A particularly interesting addition to this new AI model launch is Google bringing the Gemini 2.0 Flash Thinking Experimental model to the Gemini app. Before, you could only access it via Google AI Studio and Vertex AI, but now, as an advanced subscriber, you can follow the AI’s thought process, understand its assumptions, and see how it arrives at its conclusions.
The AI world is indeed evolving, and Google is seriously making major moves.