Connection to Ollama Turbo hosted models

[bdensmore]

Hi all,

I just signed up for Ollama's turbo offering and was hoping I could set up the ollama provider in avante to use an AVANTE_OLLAMA_API_KEY or OLLAMA_API_KEY environment variable with the endpoint pointing to https://ollama.com but I keep getting an unauthorized error when I try to use the gpt-oss:20b model in neovim. I also tried adding the API key directly in the api_key parameter with no success.

I'm able to connect to the 20b and 120b gpt-oss models with my open-webui instance using one of my generated api keys and it works great. I know the hosted turbo models are new, but I wanted to see if there is a way to hook this up or if we will have to wait for this feature to be added to the ollama provider module.

Any help would be greatly appreciated.

Thanks.