Google Gemma 3

Gemma 3 27B IT is a large multimodal instruction-tuned transformer model built using the same research foundation as Gemini models. It features a 27B parameter architecture with 62 transformer layers, 32 attention heads, and a 5,376 hidden size, optimized for advanced reasoning and instruction following. It supports a 128K token context window with RoPE scaling. The model integrates a SigLIP-based vision encoder with 27 layers and supports multimodal processing with 256 tokens per image, enabling joint understanding of text and visual inputs for complex multimodal tasks.

Gemma 3 12B IT is a multimodal instruction-tuned transformer model developed by Google as part of the Gemma 3 family built on the same research foundation as Gemini models. It features a 12B parameter architecture with 48 transformer layers, 16 attention heads, and a 3,840 hidden size, optimized for high-quality instruction following and long-context reasoning. The model integrates a SigLIP-based vision encoder with 27 vision transformer layers and supports multimodal processing with 256 tokens per image, enabling joint understanding of text and visual inputs.

Gemma 3 4B IT is an instruction-tuned multimodal transformer model built using research and technology from Gemini models. The model integrates text and vision processing, featuring a 4B parameter-class architecture with 34 transformer layers, a 2,560 hidden size, and a multimodal design combining a language backbone with a SigLIP-based vision encoder. It supports a 128K token context window with RoPE scaling for long-context understanding and can process images at 896×896 resolution encoded into 256 tokens.

Gemma 3 1B IT is a lightweight instruction-tuned dense transformer language model built using the same research and technology behind Gemini models. Built on a 1B parameter architecture, it features 26 transformer layers, 4 attention heads, and a 1,152 hidden size, using sliding-window attention mechanisms for efficient context processing. The model supports a 32K token context window and multilingual capabilities across 140+ languages, and is designed for efficient deployment on resource-constrained environments while maintaining strong performance on general language tasks.

Gemma 3 270M IT is a lightweight instruction-tuned dense transformer language model derived from the same research used for Gemini models. Built on a 270M parameter architecture, it features 18 transformer layers, 4 attention heads, and a 640 hidden size, using a combination of sliding-window and full attention mechanisms for efficient context processing. The model supports a 32K token context window and multilingual capabilities across 140+ languages, and is optimized for lightweight deployment on resource-constrained environments such as laptops and edge infrastructure.