Gemini 4 Argon

Google has released Gemini 4 Argon, the fourth version of its Gemini model, a large language model developed by the company. Gemini 4 Argon is a significant improvement over its predecessor, Gemini 3 Zephyr, with a 20% increase in scale and a 25% increase in fine-tuning data. The model has 7.5 billion parameters, a 20% increase over Gemini 3 Zephyr's 6.2 billion parameters. The model is trained on a dataset of 1.5 TB of text, which is 25% larger than the dataset used for Gemini 3 Zephyr. The improvements in Gemini 4 Argon are aimed at enhancing its conversational abilities, particularly in open-ended and multi-turn conversations. The model can be used for a wide range of applications, including customer service, content creation, and language translation.