Conversation
supportHi everyone I have a site, that i need to troubleshoot. Meow -> AI Engine is installed, and has worked fine since release, as far as I know. Now we are seeing an error code about Azure trying to find Response API URL which doesn’t exist. The deployment used is not set up with a Response API., just Chat Completions. This is what it looks like in WP: What I don’t know is where or how to fix the error I get. Apparently the model used does not support Responses API. Does anyone see/know/guess the issue here? As I said, it has worked before, and only the last couple of weeks it has been down… The page I need help with: [ log in to see the link]
Hello @marksten70 , Thanks for the screenshots. I can see a couple of issues: Missing API Key Your API Key field is empty in the environment settings. This is required for authentication, so without it, no requests will succeed. Non-standard endpoint Your endpoint (chat.uit.no) doesn’t look like a standard Azure OpenAI endpoint (those look like yourresource.openai.azure.com). Responses API mismatch In a recent update, GPT-4o Mini was tagged to use OpenAI’s newer Responses API, which changes the URL path to /v1/responses. Your endpoint likely only supports the Chat Completions path (/chat/completions), which is why it worked before and broke recently. Instead of selecting “GPT-4o Mini” from the dropdown in your deployment, manually type a custom name (e.g. gpt-4o-mini-custom). This way AI Engine won’t match the responses tag and will fall back to the Chat Completions API. I hope it helps! ☺️ Jordy.
Oh… Thank you for response. I will indeed try what you suggest. PS: I blanked out the API-key on pupose, so it didn’t show here 🙂
Hmm… I don’t seem to be able to write any custom text, like you suggest instead of choosing from drop-down menu. Am I in the right place? This reply was modified 2 months, 1 week ago by marksten70 .
@tigroumeow Just to clarify. We don’t use OpenAI as endpoint, it just looks like that. Out Uni has a locally setup based on ChatGPT – ChatUiT, which is hosted from that url. As mentioned it has been working, and your answer tells us what the problem is. I have been unable to figure out where to write in a custom model text, like you suggesed
Hello @marksten70 , Thanks for the screenshots. I can see a couple of issues: Missing API Key Your API Key field is empty in the environment settings. This is required for authentication, so without it, no requests will succeed. Non-standard endpoint Your endpoint (chat.uit.no) doesn’t look like a standard Azure OpenAI endpoint (those look like yourresource.openai.azure.com). Responses API mismatch In a recent update, GPT-4o Mini was tagged to use OpenAI’s newer Responses API, which changes the URL path to /v1/responses. Your endpoint likely only supports the Chat Completions path (/chat/completions), which is why it worked before and broke recently. Instead of selecting “GPT-4o Mini” from the dropdown in your deployment, manually type a custom name (e.g. gpt-4o-mini-custom). This way AI Engine won’t match the responses tag and will fall back to the Chat Completions API. I hope it helps! ☺️ Jordy.
Oh… Thank you for response. I will indeed try what you suggest. PS: I blanked out the API-key on pupose, so it didn’t show here 🙂
Hmm… I don’t seem to be able to write any custom text, like you suggest instead of choosing from drop-down menu. Am I in the right place? This reply was modified 2 months, 1 week ago by marksten70 .
@tigroumeow Just to clarify. We don’t use OpenAI as endpoint, it just looks like that. Out Uni has a locally setup based on ChatGPT – ChatUiT, which is hosted from that url. As mentioned it has been working, and your answer tells us what the problem is. I have been unable to figure out where to write in a custom model text, like you suggesed