Building a Flask Translation App with GLM-4-Flash API

On August 27, Zhipu released GLM-4-Flash, a free large‑model API for fast text generation, code assistance, translation, knowledge‑base Q&A, PPT assistance, mind‑map generation and more.

What is GLM-4-Flash

GLM-4-Flash is a high‑speed, free LLM API suitable for article creation, code debugging, translation, knowledge‑base Q&A, PPT assistance, mind‑map generation and more.

Typical usage scenarios

Developers can use it for data extraction (entity recognition, keyword extraction), literature reading assistance, script and model code generation, model optimization suggestions, etc. Below is a demo of a text‑to‑text translation service built with Flask.

Setup

1. Install Python 3.12.

2. Install required packages:

pip install Flask

pip install zhipuai

3. Obtain an API key from the ZhipuAI platform and set api_key = "your_key" in the code.

Running the Flask app

Execute:

python app.py

The app starts at http://127.0.0.1:5000/ where users can submit text for translation.

Key code overview

Import libraries:

from flask import Flask, request, jsonify, render_template
from zhipuai import ZhipuAI
import threading
import uuid

Initialize app and client:

app = Flask(__name__)
api_key = ""  # replace with your ZhipuAI API key
zhipuai_client = ZhipuAI(api_key=api_key)

Data structures for task management:

translation_results = {}
task_status = {}
lock = threading.Lock()

Translation function:

def translate_text(prompt, source_lang, target_lang, user_id=None):
    try:
        languages = {'en': '英文', 'zh': '中文'}
        response = zhipuai_client.chat.completions.create(
            model="glm-4-plus",
            messages=[...],
            user_id=user_id,
            top_p=0.7,
            temperature=0.95,
            max_tokens=1024,
            tools=[{"type":"web_search","web_search":{"search_result":True}}],
            stream=False,
        )
        if response and response.choices[0].message:
            return response.choices[0].message.content
        else:
            print("Error: No translated text returned.")
            return None
    except Exception as e:
        print(f"Error translating text: {e}")
        return None

Routes:

@app.route('/')
def index():
    return render_template('index.html')

@app.route('/translate', methods=['POST'])
def translate():
    data = request.json
    prompt = data.get('prompt')
    source_lang = data.get('source_lang', 'zh')
    target_lang = data.get('target_lang', 'en')
    user_id = data.get('user_id')
    if not prompt:
        return jsonify({"error": "No text provided"}), 400
    task_id = str(uuid.uuid4())
    with lock:
        task_status[task_id] = "processing"
    def handle_translation():
        translated_text = translate_text(prompt, source_lang, target_lang, user_id)
        with lock:
            translation_results[task_id] = translated_text
            task_status[task_id] = "completed"
    threading.Thread(target=handle_translation).start()
    return jsonify({"message": "Translation in progress", "task_id": task_id}), 202

@app.route('/get-translation/
', methods=['GET'])
def get_translation(task_id):
    with lock:
        status = task_status.get(task_id)
        result = translation_results.get(task_id)
    if status == "completed" and result:
        return jsonify({"task_id": task_id, "status": status, "translated_text": result}), 200
    elif status == "processing":
        return jsonify({"task_id": task_id, "status": status, "message": "Translation is still in progress"}), 202
    else:
        return jsonify({"task_id": task_id, "status": "failed", "message": "Translation failed or task does not exist"}), 404

Start the server:

if __name__ == '__main__':
    app.run(debug=True)

Users can access the home page, submit text, and retrieve translation results via the provided task ID.

In summary, GLM‑4‑Flash offers a free, high‑performance LLM API that can be quickly integrated into backend services such as this Flask translation demo.