Stream OpenAI, DeepSeek, Groq, Llama, or any compatible LLM in real-time. Features secure edge architecture and rich Markdown rendering.
Build the ultimate AI Chat experience with any model:

* AI Chatbots (SaaS).
* Internal tools using private models.
* Apps requiring high-speed inference.

This plugin is a universal streaming client designed for the "OpenAI-compatible" API standard.

Features
=======

* Universal Compatibility:
Works with any provider that supports the /chat/completions format (OpenAI, DeepSeek, Groq, Mistral, local Ollama, etc.).

* True Real-Time Streaming:
Silky-smooth "typewriter" effect using standard Server-Sent Events (SSE).

* Beautiful Markdown:
Includes a visual element that perfectly renders code blocks, tables, bold text, and lists instantly.

* Enterprise Security:
Built to work with a private Proxy / Function so your API keys are never exposed to the browser.

* Smart Buffering:
Automatically fixes fragmented network packets to prevent JSON errors common in other plugins.


Prerequisite: Security First
=====================

To ensure your API keys remain private and to enable streaming, this plugin requires a backend proxy URL (e.g., a simple Cloudflare Worker or Supabase Function).
Do not use your API keys directly in the front end.

Setup Guide
==========

1) Deploy your Function/Worker/Lambda:
Set up a simple backend function that holds your API Key (OpenAI, Groq, or DeepSeek) and forwards the request.

2) Add the Element:
Drag the "OpenAI Stream" element onto your page.

3) Add the Viewer:
Drag the "Markdown AI Viewer" element where you want the text to appear. Link its data source to the Stream AI Chat text_output / html_output.

4) Endpoint URL:
Paste the URL of your deployed proxy/worker/function.

5) Model:
Type the specific model name you want to use (e.g., gpt-4o, deepseek-chat, llama3-70b-8192).

6) Messages: Pass the user input and history. (optional)

7) Run the Workflow:
Use the action "Send Prompt".