Large language model
These models are built using deep learning techniques, particularly neural networks with many layers, which enable them to understand and produce coherent language. LLMs are trained in diverse text, encompassing books, articles, websites, and more, allowing them to learn grammar, context, facts, and nuances of human language.
One of the key characteristics of LLMs is their ability to generate text that is contextually relevant and grammatically correct. They can perform a wide range of language-related tasks, including translation, summarization, question answering, and text completion. LLMs have billions of parameters, which are the weights in the neural network that the model adjusts during training to learn language patterns.
Supported models
This table shows the LLMs that are available to use within the GenAI application:
Name | Description | Token Window | US East (Y/N) | EU (Y/N) Azure OpenAI – France Central AWS Bedrock - Frankfurt |
---|---|---|---|---|
Claude Instant (deprecating Jul 2025) | Optimized for rapid performance and competitive affordability. | 100,000 | Y | N |
Claude 2 (deprecating Jul 2025) | Extensive context window to support larger prompts, such as document inputs. | 100,000 | Y | Y |
Claude 2.1 (deprecating Jul 2025) | Claude 2 performance with additional support for hallucination reduction. | 200,000 | Y | Y |
Claude 3 Haiku | Fast and compact model for near-instant responsiveness. Can answer simple queries and requests with speed as a priority. | 200,000 | Y | N |
Claude3.5 Haiku | Fast and compact model for near-instant responsiveness. Answers simple queries and requests with considerable speed. | 200,000 | Y | N |
Claude 3 Sonnet (deprecating Jul 2025) | Ideal balance between intelligence and speed – particularly for enterprise workloads. Delivers strong performance at a lower cost with high endurance for large-scale applications. | 200,000 | N | N |
Claude 3.5 Sonnet | Twice the speed of Claude 3 Opus, ideal model for complex workflows within a cost-effective pricing. | 200,000 | Y | N |
Claude 3.7 Sonnet | Anthropic's most intelligent model to date and the first Claude model to offer extended thinking—the ability to solve complex problems with careful, step-by-step reasoning. | 128,000 | Y | N |
Claude 3 Opus (pending) | The most intelligent Claude 3 model, with best-in-market performance on highly complex tasks. | 200,000 | N | N |
Titan Lite (deprecating Aug 2025) | Designed for use-case specific generation tasks with affordability and speed as priorities. | 4,096 | Y | N |
Titan Express (deprecating Aug 2025) | High-performance model for more complex tasks involving retrieval augmented generation (RAG). | 8,192 | Y | Y |
Llama 3 8B | Ideal for limited computational power and resources, edge devices, and faster training times. | 8,192 | Y | N |
Llama 3 70B | Ideal for content creation, conversational AI, language understanding, R&D, and Enterprise applications. | 8,192 | Y | N |
Nova Micro | Text-to-text understanding foundation model. Model that is multilingual and can reason over text. | 128,000 | Y | N |
Nova Lite | A multimodal understanding foundation model that can reason over text, images, and videos. | 300,000 | Y | N |
Nova Pro | High capacity for enterprise-level computations that performs effectively in a large variety of tasks. | 300,000 | Y | N |