Where do I get a streamed response?

In the browser, open the network panel, click the request and copy the response. On the command line, run curl with -N and save the output. Both work here.

Does it call the API?

No. It only reads the text you paste. No key is needed and nothing is sent anywhere.

Are my files uploaded?

No. This tool runs entirely in your browser tab. Files are read from your disk into memory and the result is generated locally.

Is it free? Are there limits or watermarks?

Free and open source (MIT). There are no file-size caps, daily quotas, accounts or watermarks.

Does it work offline?

Yes, once the page (and any engine it needs) has loaded, it keeps working without a connection thanks to the service worker cache.

Home / AI Builder / LLM Stream Viewer

LLM Stream Viewer

Paste a streamed or plain LLM API response (OpenAI, Anthropic, Gemini, Ollama) and get the full text, reasoning, tool calls and token usage.

Free, no uploads, no sign-up. Runs 100% in your browser.

How LLM Stream Viewer works

Everything runs in your browser. Prompts, datasets and model files are never uploaded. Server sent events are split into data payloads; NDJSON and JSON arrays are read line by line. Each payload is matched to its API (Chat Completions, Responses, Anthropic Messages, Gemini or Ollama), text deltas are joined, and tool call arguments that arrive in pieces are put back together.

Input: Text, JSON, NDJSON. Output: Text.

How to use LLM Stream Viewer

Copy a response from the network panel or a curl -N run.
Paste it or drop the saved file.
Read the text, reasoning, tool calls, usage and stop reason.
Copy or download the text.

Tips

Prompts, datasets and API keys never leave your device, so these tools are safe for private work.
Tokenizers and AI models download once and are cached by your browser.

LLM Stream Viewer FAQ

Where do I get a streamed response?: In the browser, open the network panel, click the request and copy the response. On the command line, run curl with -N and save the output. Both work here.
Does it call the API?: No. It only reads the text you paste. No key is needed and nothing is sent anywhere.
Are my files uploaded?: No. This tool runs entirely in your browser tab. Files are read from your disk into memory and the result is generated locally.
Is it free? Are there limits or watermarks?: Free and open source (MIT). There are no file-size caps, daily quotas, accounts or watermarks.
Does it work offline?: Yes, once the page (and any engine it needs) has loaded, it keeps working without a connection thanks to the service worker cache.

Related tools

MCP Config Builder: Build Model Context Protocol server configs for Claude Desktop, Cursor, VS Code, Windsurf, Claude Code and Gemini CLI, or convert one app's config to another.
HAR File Viewer: Open browser network captures: waterfall, filters, headers, bodies, copy as cURL, and save a copy with secrets removed.
JSON Formatter: Prettify, validate, minify and explore JSON as a tree.
Token Counter: Count tokens with the real GPT-4o, GPT-4, Llama 3, Mistral, Qwen, Gemma or DeepSeek tokenizer, see each token, context use and cost.