Also note that the AI hasn’t been trained on coding for the latest AI models and API methods. That AI model you show no longer exists. You’ll have to resort to reading the API Reference, and learning from parameter descriptions and examples shown there to write your own code.
This is more a “text processing batch process” than a “chat history” task, so you can make a for loop that substitutes different languages from a list into your “translate the user’s text into {output_language}” instruction. I just did that as an example.
import openai
client = openai.OpenAI()
translated_outputs = [] # Python list to store the translated AI responses
messages = [
{"role": "system", "content":
'You are an AI-based language translator. Output will be only a full translation of user text. '
'Preserve formatting. Preserve code.'},
{"role": "system", "content": 'output language placeholder'}, # list item 1
# user provides the text in original language
{"role": "user", "content": 'The function `get_weather` retrieves the local forecast from an API.'},
]
for language in ["French", "Spanish", "Indonesian",]:
messages[1]['content'] = f"As response, translate input into {language} language."
response = client.chat.completions.create(
messages=messages, model="gpt-3.5-turbo", top_p=0.1, max_tokens=900,
)
translation_format = {"language": language, "content": response.choices[0].message.content}
translated_outputs.append(translation_format)
print(translated_outputs) # whatever you want to do with list of translations
this produces a python list with python dictionaries:
[
{‘language’: ‘French’, ‘content’: “La fonction get_weather récupère les prévisions locales à partir d’une API.”},
{‘language’: ‘Spanish’, ‘content’: ‘La función get_weather recupera el pronóstico local de un API.’},
{‘language’: ‘Indonesian’, ‘content’: ‘Fungsi get_weather mengambil ramalan cuaca lokal dari sebuah API.’}
]
Having the AI produce multiple outputs at once from one instruction and one input text is also possible, but the quality will start to degrade and be limited around 800 tokens total.