I recently completed fine-tuning two models (GPT-3.5 and GPT-4) using OpenAI’s fine-tuning service. Both fine-tuning jobs were successful, and I received the status: succeeded for both jobs, along with the respective model IDs. However, I am encountering a persistent issue when trying to use these models via the API or the Playground. When I attempt to use the fine-tuned models in the playground, they are not listed in the model dropdown menu. Instead,I see The model ft:gpt-3.5-turbo-0125:********::<model-id> is not available. The link you opened specified a model that isn’t available for your org. We’re using the default model instead.
Look to the very upper-left of the platform site. It has an organization selection and a project selection within them. The organization must match you and the API key organization that was used to create the fine-tuning. If a member of multiple orgs, t is possible to set default keys to bill to a different organization, or to change the organization or project by forgotten environment variables.
It would be odd if the same environment variables and system that submitted the fine tuning can’t also submit inference calls. You did use environment variables, right? You are using an API key and project ID that didn’t already have model and endpoint limitations, that didnt’ take on the new model?