Tried hosting the zeta as well and it is giving me 400 error code

[ShivamRawat0l]

Tried hosting it online but there was log of delay which is on my part need to find better server but the things it suggested were plain wrong. It suggested me to import preact which i am not even using.

[leonardcser]

Could you share a bit more what service/platform you tried/used?

[ShivamRawat0l]

Yeah as I am based in India I tried hosting zeta 2 which you had provided in A30 GPU 24GB but it was pretty slow.
I guess the latency from the server was not much it was the code completion only that was taking time. I asked gemini and it said that this device was powerful enough for this model. I could choose a better device as well and test it with that but it wont be sustainable for me.
Therefore i hosted the qwen 1.5 model locally and the instant results are something i liked a lot but supermaven had similar results and it was similarly fast.
Maybe hosting a zeta 2.1 and giving it as a subscription if there are a lot of users could be done otherwise hosting for a single person these models are not worth it.

Also would like to say like sometimes i get ghost text and sometimes suggestions are in normal text. Like we can color code addition and delete by default as well i think like it is done in cursor. Just a suggestion.

Let me know if there is anything else.