I am using the openai api to define pre-defined colors and themes in my images. However, I get returns stating that the model is not capable of viewing images. I got this to work with 3.5 but tried with gpt-4o and cannot get it to work. Am I using the wrong model or is the API not capable of vision yet?
2 Likes
Welcome to the community!
Vision is not available on 01 model… yet. It’s supposed to be coming, I believe. Lack of system message and other things are missing too.
Stay tuned…
ETA: Oops you asked about 4o which should be available… hrm…
https://platform.openai.com/docs/guides/vision
What error are you getting? What usage tier are you?
https://platform.openai.com/docs/guides/rate-limits/usage-tiers?context=tier-two
1 Like
Welcome @emolitor.cota
I’d strongly recommend reading the vision capability guide.
Currently you can consume vision capability gpt-4o , gpt-4o-mini or gpt-4-turbo.
1 Like
Guys I believe it was just gaslighting me. This was using 4o. It insisted that it did not have capabilities to analyze images. I ended up changing my prompt and it started working.
2 Likes
Yeah, prompt is still important. I’ve seen ChatGPT claim it can’t do images, etc before…
Glad you got it sorted! Thanks for coming back to let us know.
1 Like
Yeah, I noticed That happened to me today as well. I found myself reposting my screenshots and asking it again and it finally recognized.
1 Like