So had a fun problem show up recently boundary #questions

So had a fun problem show up recently

jeremy.london

09/28/2026, 9:30 PM

So had a fun problem show up recently We have azure-openai and it only support the deployment flow url and it worked fine.. But newer openai models use the responses endpoint So was looking at the docs and it says to use openai-generic but is that /chat/completions or /responses compatible? https://docs.boundaryml.com/ref/llm-client-providers/microsoft-foundry So im guessing i need to setup the responses client in my registry https://docs.boundaryml.com/ref/llm-client-providers/open-ai-responses-api But docs are not super clear so just asking here if there is any guidance https://cdn.discordapp.com/attachments/1253172394345107466/1554244046367756391/image_4.png?ex=6abc2e11&is=6abadc91&hm=adc90648995f71f94d6618caa11ffbc4d25c10dff8a7b70511c669541bcd4945& https://cdn.discordapp.com/attachments/1253172394345107466/1554244047210807307/image_3.png?ex=6abc2e11&is=6abadc91&hm=1fb13d6dd5444f355f5dc0c550ffbfea67f31e05a2d7d9e75ea2d495ad548512&

jeremy.london

09/29/2026, 1:37 AM

By the way i added

openai-responses

to my client registry and got this working. Its silly Azure has different ways based on different models. but all is working again now. Docs could use a little tune up for this provider. If they want to use the openai-generic and their url supports /chat/completions then can do so, but if their url is a responses url they need to use openai-responses - So Azure has /deployment urls, and openai /chat/completions - But Azure also has only /responses From what I see 5.6 and above models only do the responses flow, but older models use the older deployment or openai chat api for serving. And microsoft just never documented this anywhere 🙃

Vaibhav (BAML 🐑)

09/29/2026, 4:00 PM

Hi @jeremy.london thanks for self resolving this!

baml-sam

09/29/2026, 6:17 PM

The

/responses

switch is really annoying at an ecosystem level - pretty much all the third-party inference providers standardized on the chat completions API and

/responses

is slowly happening as everyone gets pulled in as they pick up the new models

jeremy.london

09/30/2026, 12:48 AM

Yeah i was putting off putting the responses support in because all our docs focus around find a openai-compatiable chat completion provider and plug it in All in all it took around 5 minutes to go wire up and add responses - a little url parsing to handle the different ways the endpoint might come in. Love BAML so much.. best tech stack choice - been in place since May 2024 and this was the first real gotcha I hit. Killer guys!