Google Unveils Advanced Gemma 2 AI Models for Developers

The new 9B and 27B parameter models offer high performance and efficiency, accessible through various AI frameworks.

Gemma 2 models outperform larger competitors in performance and efficiency tests.
The 27B model handles complex tasks with greater accuracy and depth, while the 9B model is optimized for speed and computational efficiency.
Both models feature an 8192 context length and use Rotary Position Embeddings for better handling of long sequences.
Developers can integrate Gemma 2 with popular AI frameworks like Hugging Face Transformers, JAX, PyTorch, and TensorFlow.
Google offers Gemma 2 free of charge for academic research and provides credits for first-time Google Cloud customers.