New GPT-4 Turbo with Vision model

API
,
1

The promotional e-mail for the new Vision model confused me. The mail contains the following sentence:
“It is now generally available (GA) in the API under the gpt-4-turbo alias.”
I thought Vision could be used with the Assistant API. When I tested it via Playground, it was still unusable.
The interesting thing is that when I add a PNG or JPEG file to the thread and send a request, an error occurs during the run and I get an error every time. In the past, this would not happen and the answer would be that I cannot access the files.

Vision was opened with the assistant api and is there an error or is it still only available with the completion api? Does anyone have an idea?

1 Like
2

The promotional material was rather confusing, but as stated in the documentation the assistants api still does not support image inputs at this time.

Bonus:

Plese note that the Assistants API does not currently support image inputs.

Typo spotted!