Inconsistent response from OpenAI Deployments API

Question

According to the documentation, the Azure OpenAI Deployments API should return token limits for the deployment (if applicable).

    "capabilities": {
      "area": "EUR",
      "chatCompletion": "true",
      "jsonObjectResponse": "true",
      "maxContextToken": "128000",
      "maxOutputToken": "16834",
      "assistants": "true"
    },

However, it seems in some cases, this information is not included. At least in my case, I don't get the information for gpt-4o or gpt-4.1, but it works for gpt 3.5, gpt-4, gpt-4-vision, gpt-4o-mini.

Answer 1

Hi there Taylor Nelson

Thanksn for using QandA platform

Yup i hve seen it too, OpenAI Deployments API is supposed to return maxContextToken and maxOutputToken, some deployments like gpt-4o or gpt-4.1 may not include those fields, even though they’re present for gpt-3.5 or gpt-4. likely due to backend deployment schema differences hope this will be addresred in the next rollout

The best workaround is to refer to the official model documentation for token limits or test them programmatically.

if this helps kindly accept the answer thanks much.