Salve,
Sto cercando di utilizzare il modello DeepSeek-V3 tramite le API di Hugging Face per un progetto personale, senza scaricarlo in locale. Tuttavia, sto riscontrando l’errore:
{“error”:“Task not found for this model”}
Potreste indicarmi se esiste un task preconfigurato o un endpoint API per questo modello?
Grazie in anticipo per l’aiuto!
It’s not possible because the Serverless Inference API is turned off…
You can tell by the color and state of the 
mark.
And the Endpoint API is a paid service…
many thanks for your answer.
would things be better if I decided to download the model and use it locally?
if so, how much disk space do I need?
I was surprised when I looked at the repository to see how big it was. The size is far over 500GB…
I doubt there are many people who have that much RAM, let alone VRAM…
So it’s a waste of time to try, so let’s look for an online service. It seems there is an official one.
Some brave souls are running it locally!
do you refer to this one?
https://platform.deepseek.com/
if so, do you think it is possible to use it for free? i’m newbie on AI and it seems that any kind of solution goes into paid service 
do you think it is possible to use it for free?
idk…Because the hardware required to do this properly cannot be purchased for $10,000. Even if it could be used for free, the number of times it could be used would be quite limited…
i have the same feeling. so i try to have an hint by you instead of by AI 
the original project i’m trying to get:
https://digitalgarden-puce-rho.vercel.app/01-published-articles/home-voice-assistant/
uses openai api. but i experienced that i must input my credit card data to use openai api even if some (how many?) free credits are available. and i don’t want this.
so i’m looking for openai api alternatives for free, used as a remote service because i have very, very low resources (the project indicates a RPI 5 but i’m using a TV box MXQ Pro 4K with SOC S905W 2GB RAM and 32GB microSD storage).
do you think this is an achievable goal or a huge waste of time?