LLM Stream Debugger
Send OpenAI or Claude streaming requests and inspect events, assembled replies, tool calls and request timing.
FreeBrowserAbout LLM Stream Debugger
Send a streaming request to an OpenAI-compatible or supported Claude endpoint and inspect events as they arrive. Compare raw chunks, reconstructed text and timing when debugging a model integration.
When to use this tool
- Diagnose malformed or interrupted model streams.
- Compare incoming chunks with the final displayed response.
Limits to know before using it
Browser CORS restrictions can block a provider. Protocol compatibility and usage fields vary by endpoint; the external provider can charge for the request.
Data handling and cost
Requests and the API key go directly from your browser to the endpoint you configure. The key stays in page memory rather than persistent storage. Use a trusted endpoint and clear the key after testing.
The debugger itself is free. The model provider can charge for requests sent to its endpoint.
Frequently Asked Questions
Only in page memory, never persisted. It is sent directly to the endpoint you enter. Use a trusted endpoint and clear your key after testing.
The endpoint must allow CORS from this website. HTTPS pages also require HTTPS endpoints. Check endpoint, credentials, network access and provider permissions.
No. It displays tool call names and streamed arguments. Calls are not executed and no follow-up messages are sent. Provider usage charges may apply.