Deploy a model

Click Deploy on a completed job to turn its output model into one you can use in the Playground.

When a job finishes successfully, its detail view shows a Deploy button. Click it, and the trained output model becomes a running model you can pick in the Playground like any other.

1. 1
   Open the completed job

   From the Jobs list, click into a job that finished successfully.
2. 2
   Click Deploy

   The console provisions the output model and adds it to your model list.
3. 3
   Use it in the Playground

   Pick your fine-tuned model from the model dropdown and chat with it like any other model.

Manage deployments

The Deployments tab lists every fine-tuned model you've deployed and its status. Click Stop on a deployment to free the GPU it's holding when you're done using it.

Download adapter weights

For LoRA and QLoRA output models, the job's detail view includes a Download weights button to save the adapter weights locally if you want to run the model yourself elsewhere.