AI Models Guide
DataQI supports Azure-hosted large language models (LLMs). Each assistant within the platform can be assigned the most suitable model for its purpose, balancing accuracy, speed, and cost.
Note: Model availability depends on your organization’s Azure configuration. Not all models may be enabled for your instance. Contact your administrator or the DataQI team if you need access to specific models.
Azure-Available Models
| Provider | Model | Accuracy | Speed | Cost | Notes |
|---|---|---|---|---|---|
| OpenAI | GPT-5 | Very High | Medium | Very High | Latest with 272k context |
| GPT-5 Mini | High | High | Medium | Real-time reasoning | |
| GPT-4.1 | High | High | High | Enhanced GPT-4 | |
| GPT-4.1 mini | High | Medium | Medium | Enhanced GPT-4 | |
| GPT-4o Mini | Medium | High | Low | Default model | |
| Meta | Llama 3.1 70B | High | Medium | Medium | Latest large model |
| Llama 3.1 8B | Medium | High | Low | Balanced performance | |
| Llama 3.1 405B | Very High | Low | Very High | Largest available | |
| Mistral | Mistral Large | High | Medium | Medium | Premium model |
| Mixtral 8x7B | Medium | High | Low | Mixture of experts | |
| xAI | Grok-2 | High | Medium | High | Latest Grok model |
| Grok-2 Mini | Medium | High | Low | Lightweight Grok |
Model Selection
DataQI administrators can select the model used by each assistant based on the use case and priorities (e.g., document summarisation, search, real-time chat). Models can be changed as needs evolve.
If your organisation requires a specific model or deployment pattern, please contact the DataQI team for guidance and provisioning support.
Last updated on