#@louisjordan It seems the bug is only
1 messages · Page 1 of 1 (latest)
Let me know if we can be of service
Our team has just deployed a fix. Can you confirm?
Yes, works now. Can I ask if the issue was our side or 11l side? ( just so we can avoid if it was us )
Thank you for the feedback, we have gotten orders already for a demo build on this, it just needs to be bulletproof 🙂
I started seeing this same issue last night, and it still isn't resolved for me. Is the fix not fully roled out yet?
when I use any LLM other than custom it works fine, but when I choose custom LLM it never reaches out to the URL, never says the first message, and the websocket to the agent seems to just disconnect as soon as it connects.
Hmm sorry about that Robert
Could you give me some screenshots so I can understand the issue a little better? Agent ID + settings configuration
so if I just create a template of Eric the support rep it works. It works with any LLM other than custom, but if I switch custom then the websocket gets closed immediately upon connect and the custom URL is never touched. You can see an example with Agent ID dqOfDN3ppqFWIZxtCUOE
0 change to the config other than choosing Custom LLM, putting in my URL, and typing "model" or "custom" or "gpt4o" (it doesn't matter what) into the "Model ID" field
when I do "Test AI Agent" in the UI, my developer console starts scrolling with "WebSocket is already in CLOSING or CLOSED state." repeatedly
if I change back from custom LLM to any other LLM and then do the "Test AI Agent", those errors do not happen and the conversation works properly.
Just seeing this, I came here due to the same issue. I tested it with two agents, e.g. 8hvXEe5t9acwapTTJWtn
I texted more details here #📟│agents-chat message
Ok just tested https://api.openai.com/v1/chat/completions and https://api.openai.com/v1 as a custom LLM but that also does not work 😦
Hey @cinder skiff & @bright smelt - I'm really sorry this is happening - let me check this and get back to you.
I have just submitted an internal report - will get back to you once this has been resolved
Thanks for your patience both
Thank you so much
@cinder skiff @proven scroll seems like they added a parameter for custom LLMs #📟│agents-chat message
Thank you! I'll give this a try.
it's working.