Google’s Gemma AI models have surpassed 150 million downloads

Gemma AI models hit 150M downloads with 70K variants on Hugging Face, yet trail Meta's Llama in total downloads.

: Google's Gemma AI models, launched in February 2024, have surpassed 150 million downloads worldwide. This collection of AI models also spurred the creation of more than 70,000 variants on the AI developer platform Hugging Face, as announced by Omar Sanseviero from Google DeepMind. Despite these impressive numbers, Gemma still lags behind Meta's Llama models, which have achieved over 1.2 billion downloads by late April 2025. Both Gemma and Llama face criticism over their custom, non-standard licensing terms, raising concerns over their commercial use.

Google's Gemma AI models have reached a significant milestone with over 150 million downloads, marking a substantial achievement for the company since the models were launched in February 2024. These models were created to rival other open model families such as Meta's Llama. Omar Sanseviero, a developer relations engineer at Google DeepMind, announced this milestone on X. The Gemma models are designed to be multimodal, supporting over 100 languages and processing both images and text, enhancing their versatility and applicability in varying fields.

Despite celebrating over 150 million downloads, Google's Gemma is still outpaced by Meta's Llama, which has reportedly been downloaded over 1.2 billion times by April 2025. This stark contrast highlights the competitive landscape of AI model downloads, where Meta holds a clear lead in terms of quantity. The Gemma models, however, are notable for the 70,000 variants created on the platform Hugging Face, showing a robust level of engagement from developers.

Gemma's functionality extends to particular applications, with models fine-tuned for specific domains like drug discovery, suggesting a targeted approach in its development strategy. This has opened doors for the AI models to be used in specialized fields, potentially increasing the models' utility and relevance across diverse technological landscapes. Additionally, the models' ability to identify emotions is causing a stir among experts concerned about ethical implications.

However, both Gemma and Llama are under scrutiny for their non-standard licensing terms. These terms have been a point of contention, with developers expressing concerns over potential risks when using these models commercially. This highlights the ongoing issues within the AI community regarding the balance between open innovation and restrictive legal frameworks.

The popularity of these AI models ties into the broader trend of generative AI and machine learning technologies gaining traction across various industries. As organizations continue to integrate these technologies, the demand for more accessible and customizable models is likely to increase. The engagement and feedback from the developer community, particularly regarding future versions of Gemma, will play a crucial role in shaping the path forward for these models and their acceptance in the industry.

Sources: X, TechCrunch

AIGenerative AIMachine Learning