Please add moonshotai/Kimi-K2.6 · Hugging Face, seems to be a great model.
Replace kimi k2 instruct with Kimi k2.6 model, and keep kimi k2 thinking model, it is a good model
Add Kimi K2.6 or Replace Kimi K2 Instruct (not the Kimi-K2-Instruct-0905 model, the other one) with Kimi K2.6.
It is really good at coding, and building ui. Could you please add it?
Yeah i think they will eventually add Kimi K2.6 to nvidia nim
They added GLM5.1 which has a much higher cost per million token, so…
Let’s go! Thanks for the great support… we need this model soon!
good news guys!!! they added deepseek-v4-pro ( https://build.nvidia.com/deepseek-ai/deepseek-v4-pro ) which is even better than Kimi K2.6 at coding, and it has 1M context
good news again guys!!!
it seems like they added kimi k2.6 to the api url
despite it not being available on https://build.nvidia.com/moonshotai/kimi-k2.6
it is available on the API URL
https://integrate.api.nvidia.com/v1/models
I have the same problem no matter what settings I use I go through the chat or the API. It’s very random, regardless of the type of prompt. I also have a lot of trouble using many models like Kimi k2.6, Deepseek-V4, etc. The connection is often reset by the server, or sometimes I have to wait more than half an hour. It feels like since Kimi 2.6 arrived, the servers have been overloaded.