Google has launched Gemma 3, an update to its AI models, aimed at developers creating applications across different hardware. The model supports over 35 languages and can process text, images, and videos. Google asserts that it outperforms competitors like Facebook's Llama and OpenAI on single GPU setups, emphasizing lowered hardware requirements. While promoting Gemma via cloud credits for developers, Google also addressed concerns regarding potential misuse, indicating a low risk level despite the model's advanced capabilities and the ongoing debate over its open-source status due to licensing limitations.
The launch of Gemma 3 marks Google's commitment to providing advanced AI tools for developers, with significant performance claims over its competitors, tailored for various hardware.
Google claims Gemma 3 is the 'world's best single-accelerator model', designed for efficient use across devices and providing support for over 35 languages.
Despite enhanced capabilities, Google assessed Gemma 3 for potential misuse, concluding it poses a low risk for harmful applications, a pivotal aspect of its release.
The release of Gemma 3 highlights the ongoing discussion about what defines an 'open' AI model, especially in relation to Google's licensing restrictions.
Collection
[
|
...
]