run.py

import os

import gradio as gr
from openai import OpenAI

def chat(prompt: str, oauth_token: gr.OAuthToken | None = None) -> str:
    """Send a prompt to GLM-5.3-Flash via HF Inference Providers and return the reply."""
    api_key = oauth_token.token if oauth_token else os.environ.get("HF_TOKEN")
    if not api_key:
        return "🔒 Please sign in with Hugging Face (top of the page) to run this workflow."
    client = OpenAI(
        base_url="https://router.huggingface.co/v1",
        api_key=api_key,
    )
    stream = client.chat.completions.create(
        model="zai-org/GLM-5.3-Flash:fastest",
        messages=[{"role": "user", "content": prompt}],
        stream=True,
        max_tokens=8192,
    )
    out = []
    reasoning = []
    for chunk in stream:
        if not chunk.choices:
            continue
        delta = chunk.choices[0].delta
        if getattr(delta, "reasoning_content", None):
            reasoning.append(delta.reasoning_content)
        if delta.content:
            out.append(delta.content)
    answer = "".join(out).strip()
    if answer:
        return answer
    if reasoning:
        return "⚠️ The model returned only reasoning (no final answer):\n\n" + "".join(reasoning)
    return "⚠️ The model returned an empty response. Please try again."


demo = gr.Workflow(bind={"glm_chat": chat})

if __name__ == "__main__":
    demo.launch()