What makes Amplify GenAI Different
Amplify GenAI gives you access to multiple models within one, secure location. Models can be utilized between conversations and should keep their internal chat memory – or contextual window. Choosing a model, or swapping models to continue working through your project and conversational needs can ensure that your model always aligns its expertise with your project, idea, or conversational needs. Testing each model with various prompts helps identify their strengths and weaknesses.
Model Capabilities
Note: Currently, there is no way for any of the models within Amplify – regardless of their abilities outside the Amplify interface – to web search and find sources to cite the outputs you receive. The models will only be able to pull information up to the date that they were last trained, and all outputs should be verified independently for accuracy.
(Vanderbilt Documentation)
Each model has unique strengths that affect the content, tone, and style of the output. For example, one model might excel at technical writing, while another is better for creative storytelling.
- ChatGPT 3.5 Consider for simpler queries where the latest information isn’t crucial. Performs well with precise tasks, discussing general topics, brainstorming, drafting questions, organizing fundamental information, proposing innovative ideas, and offering recommendations. Trained on information available through January 2022.
- ChatGPT 4-Turbo This is an omni model which can be used for complex tasks requiring advanced understanding. Can carry out complex mathematical operations, code assistance, analyze intricate documents and datasets, demonstrated critical thinking, in-depth context understand and supports images as input. Trained on information available through October 2023.
- ChatGPT 4o Consider for complex tasks requiring advanced understanding. Offers further advanced intelligence over its predecessors. Can carry out complex mathematical operations, code assistance, analyze intricate documents and datasets, demonstrates critical thinking, in-depth context understanding, and supports images as input. Trained on information available through April 2023.
- Claude 3-5 Sonnet Consider for advanced tasks with the most up-to-date information. Can write, edit and execute code with sophisticated reasoning, understands user context, offers orchestrating multi-step workflows, can navigate unstructured data, write creatively, understands nuance and humor and supports images as input. Trained on information available through April 2024
- Claude 3 Haiku Consider for high-velocity tasks with near-instant responsiveness and emphasis on security and robustness through minimized risk of harmful outputs. Best for simple queries, lightweight conversation, rapid analysis of large volumes of data, handling of much longer prompts, and supports images as input. Trained on information available through August 2023.
- Mistral 7b Instruct Consider for pre-filtering or high-volume tasks. Best for text-related tasks such as summarization, classification, answering questions, creative content generation, and light dialogue. Trained on information available through September 2023.
- Mistral Large Consider for complex tasks and advanced understanding without the need for recent knowledge. Offer a greater level of intelligence compared to its predecessors. Excels in complex reasoning, text understanding, transformation, code generation, and offers advanced capabilities for multilingual reasoning and analysis. Trained on information available through 2021.
- Mistral 8x7b Instruct Consider for rapid processing and task-specific fine-tuning. Best for text summarization, questions answering, text classification, code generation and creative content generation. Trained on information available through September 2023.