Large language models offer capabilities that overlap with more traditional approaches to natural language processing tasks like translation. Multimodal large language models have an even broader overlap with traditional speech and image models. You can now choose which approach best suits your needs. Here, you will learn about their strengths and weaknesses compared to neural machine translation techniques and receive an overview of the latest advancements in our SOTA Cloud Translation API, combing ease of use with contextual capabilities of generative AI models, to enhance our customers' translations at scale. Experience how new models trained to perform both transcription and translation can go from speech to text in a target language using one large model.

Click the blue “Learn more” button above to tap into special offers designed to help you implement what you are learning at Google Cloud Next 25.