AI Studio / Deploy

Deploy Gemma 4 26B A4B

Mixture-of-experts model with vision input, about 4B parameters active per token. Served with Ollama on a GPU you reserve on the Nosana network; you pay from your own wallet and get a private endpoint with a chat playground.

Expect 3 to 10 minutes after a node starts the job before the model answers: the node downloads the image and the weights first, and that time counts toward your reservation.