Supported Language Models
UsageGuard supports a wide range of Language Models (LLMs) from various providers, allowing you to leverage the power of different AI models while maintaining a consistent API and robust safeguards.
Overview
UsageGuard acts as a proxy for multiple LLM providers, offering a unified interface for interacting with various models. This page provides an overview of the supported LLMs, their capabilities, and any provider-specific considerations.
The list of supported models is continuously expanding. Check our changelog for the most recent additions.
Supported Providers and Models
UsageGuard offers a unified inference API for various language models.
| Name | Model Id | Capabilities (Input) | Available | Description |
|---|---|---|---|---|
| Mistral Large | mistral.mistral-large-2402-v1:0 | text | Yes | Large language model by Mistral AI. For advanced NLP tasks and complex reasoning. |
| Mistral Small | mistral.mistral-small-2402-v1:0 | text | Yes | Efficient small model by Mistral AI. For lightweight NLP tasks and quick responses. |
| Amazon Titan Text G1 - Express | amazon.titan-text-express-v1 | text | Yes | Fast and efficient text model by Amazon. Ideal for rapid text generation and real-time applications. |
| Amazon Titan Text G1 - Lite | amazon.titan-text-lite-v1 | text | Yes | Lightweight text model by Amazon. Suitable for efficient text processing and mobile applications. |
| Amazon Titan Text G1 - Premier | amazon.titan-text-premier-v1:0 | text | Yes | Premium text model by Amazon. For high-quality text generation and complex NLP tasks. |
| Meta Llama 3.2 1B Instruct | us.meta.llama3-2-1b-instruct-v1:0 | text | Yes | Compact instruction-tuned model. Efficient for instruction following and quick responses. |
| Meta Llama 3.2 3B Instruct | us.meta.llama3-2-3b-instruct-v1:0 | text | Yes | Larger instruction-tuned model. For more complex instruction tasks with better quality. |
| Mistral 8x7b Instruct | mistral.mixtral-8x7b-instruct-v0:1 | text | Yes | Large language model for instruction. Suitable for complex instruction-based tasks and educational tools. |
| Mistral 7b Instruct | mistral.mistral-7b-instruct-v0:2 | text | Yes | Smaller instruction-optimized language model. For lightweight educational tools and guided tasks. |
| Meta Llama3 70b Instruct | meta.llama3-70b-instruct-v1:0 | text | Yes | High-capacity language model for instruction. Ideal for advanced educational platforms and detailed guidance. |
| Meta Llama3 8b Instruct | meta.llama3-8b-instruct-v1:0 | text | Yes | Efficient language model for instruction. Suitable for general instruction and lightweight educational tools. |
| Anthropic Claude 3 Opus | anthropic.claude-3-opus-20240229-v1:0 | text,image,document | Yes | Comprehensive language model. For a broad range of tasks and versatile applications. |
| Anthropic Claude 3 Haiku | anthropic.claude-3-haiku-20240307-v1:0 | text,image,document | Yes | Specialized for concise and poetic outputs. Ideal for creative writing and poetry generation. |
| Anthropic Claude 3.5 Sonnet | anthropic.claude-3-5-sonnet-20240620-v1:0 | text,image,document | Yes | Advanced poetic and structured language model. For high-quality structured writing and sonnet creation. |
| Anthropic Claude 3 Sonnet | anthropic.claude-3-sonnet-20240229-v1:0 | text,image,document | Yes | Poetic and structured language model. Suitable for poetic tasks and structured creative writing. |
| Anthropic Claude v2 | anthropic.claude-v2:1 | text | No | General purpose language model. For versatile applications and general language tasks. |
| Open AI GPT 3.5 Turbo | gpt-3.5-turbo-0125 | text | Yes | General purpose language model. Ideal for chatbots, content creation, and language tasks. |
| Open AI GPT-4o | gpt-4o-2024-05-13 | text,image | Yes | Advanced language understanding and generation. For complex tasks and detailed content creation. |
| Open AI GPT-4o-mini | gpt-4o-mini-2024-07-18 | text,image | Yes | Cost effective, lightweight language model. Suitable for lightweight tasks, chatbots, and content creation. |
| Open AI GPT 3.5 Turbo Instruct | gpt-3.5-turbo-instruct | text | Yes | Instruction-optimized language model. For educational tools and guided instructions. |
Switching Between Models
One of the key benefits of UsageGuard is the ability to easily switch between different LLMs without changing your application code. To switch models:
- Create a new or edit existing connection
- Enable the new model(s) to the connection
- Update your API calls to use the new connection ID (if new connection, add
x-connection-idheader) - UsageGuard will handle the rest, including any necessary request transformations
Newly released models are added automatically to your existing connection, you can start using them immediately by sending the new model id in your request.
You can alwayes choose to disable a model from a connection, this will prevent it from being used in your requests.
Best Practices
- Model Selection: Choose the appropriate model based on your specific use case and performance requirements.
- Cost Management: Monitor your usage and leverage UsageGuard's cost control features to manage expenses.
- Content Policies: Be aware of each provider's content policies and use UsageGuard's moderation features to ensure compliance.
- Performance Optimization: Use model-specific best practices for prompt engineering and request formatting to get the best results.
Troubleshooting
If you encounter any issues with a specific model or provider:
- Check the Status Page for any known issues or outages
- Review the API Reference to learn more about the model's specific parameters.
- Consult our Error Handling Guide for common issues and solutions
If you need further assistance, don't hesitate to contact our support team.
Next Steps
Now that you're familiar with the supported LLMs, you're ready to start leveraging these powerful models in your applications: