Amazon's most capable multimodal model for complex reasoning tasks. Processes text and images with up to 1M context window.