NVIDIA API Client

A plain vanilla HTML/JS/CSS web interface for making requests to the NVIDIA API chat completions endpoint.

• Form Interface: Clean, user-friendly form with all NVIDIA API parameters
• API Key Storage: Securely stores API key in browser cookies for convenience
• Request Persistence: Remembers last request parameters using localStorage
• Streaming Support: Handles both streaming and non-streaming responses
• Real-time Response Display: Shows API responses in real-time with proper formatting
• Error Handling: Comprehensive error messages with detailed diagnostic information, troubleshooting guidance, and full request/response context

The interface includes all parameters from the original curl command:

• API Key: Your NVIDIA API key (stored in cookies)
• Model: AI model to use (default: openai/gpt-oss-20b)
• Messages: JSON array of conversation messages
• Temperature: Controls randomness (0-2, default: 1)
• Top P: Controls diversity (0-1, default: 1)
• Frequency Penalty: Reduces repetition (-2 to 2, default: 0)
• Presence Penalty: Encourages new topics (-2 to 2, default: 0)
• Max Tokens: Maximum response length (default: 4096)
• Stream: Enable streaming responses (checkbox, default: checked)
• Reasoning Effort: Model reasoning level (low/medium/high, default: medium)

1. Enter your NVIDIA API key (will be saved for future use)
2. Modify any parameters as needed
3. Click "Send Request" to make the API call
4. View the response in the right panel

• Built with vanilla HTML, CSS, and JavaScript
• No external dependencies or frameworks
• Responsive two-column layout
• Proper error handling and loading states
• Cookie-based API key persistence
• localStorage for form state persistence

Makes requests to: https://integrate.api.nvidia.com/v1/chat/completions

Equivalent to the provided curl command with all the same headers and data structure.