# AI Runtime Security Proxy API Format Support

Runtime Security supports the following endpoints and formats.

| Endpoint | Format | Supported | Tool use | Images | Streaming |
|  --- | --- | --- | --- | --- | --- |
| **OpenAI** | Chat Completions |  |  |  |  |
| **Anthropic** | Messages |  |  |  |  |
| **Gemini** | Chat completion |  |  |  |  |
| **Azure Foundry** | Chat completion (OpenAI inference) |  |  |  |  |
| **AWS Bedrock** |  |  |  |  |  |
| Mistral | Text completion |  |  |  |  |
| Llama | Text completion |  |  |  |  |
| Anthropic | Anthropic messages API |  |  |  |  |
| **AWS Sagemaker** | Sagemaker payload |  |  |  |  |
| **Databricks** | Foundation model API |  |  |  |  |