Google's open artificial intelligence model Gemma 2 has been released.
Google announced the release of the Gemma 2 open artificial intelligence model family. For Gemma 2, which was introduced a few months ago, Google is entering the competition with two different model sizes.
At the Google I/O 2024 developer conference held last May, Google shared some updates for Gemma, a family of open (but not open-source) models comparable to Meta’s Llama and Mistral’s open models, and announced that the new version would be released in June. Now, the software giant has announced the availability of the Gemma 2 family.
Gemma 2 is now available
Google announced that the open lightweight model series Gemma 2 will be available to researchers and developers via Vertex AI starting next month. The company had already announced that there would be two different sizes, but had only mentioned the 27-billion-parameter model previously. In the new announcement, it was revealed that a 9-billion-parameter model would also be offered.
The standard Gemma models have so far only been available in versions with 2 billion and 7 billion parameters. Therefore, the new 27-billion-parameter model is a significant step forward. The new Gemma 2 is optimized to run on Nvidia’s next-generation GPUs, a single Google Cloud TPU host, and the Vertex AI service. These models are aimed at developers who want to add artificial intelligence to consumer-focused devices such as applications, smartphones, IoT devices, and computers.
For developers working on device-based tools, the 9B model is recommended, while the 27B model is more suited for cloud-based and more intensive AI workloads. These models are open-source, making them easily customizable and integratable into various projects that Google might not have anticipated.
The 27B model of Gemma 2 has already been added to Google AI Studio. Therefore, you can directly test the capabilities of the new model. Additionally, you can make comparisons with Gemini 1.5 and 1.0 versions here (model selection is done from the side panel). Lastly, Google announced that they will release a third model with 2.6B parameters, which will be lighter but still powerful, to be included in the Gemma 2 family.
