apiVersion mismatch of Azure o4-mini & Azure o3-mini

[LayerFort]

config.yaml

name: GPT-o3-mini
  provider: azure
  model: o3-mini
  apiKey: #
  apiBase: https://#.cognitiveservices.azure.com/
  engine: o3-mini
  apiVersion: "2024-12-01-preview"

Dev tools

Error: HTTP 400 BadRequest from https://#.cognitiveservices.azure.com/openai/deployments/o3-mini/chat/completions?api-version=2023-07-01-preview

{"error":{"code":"BadRequest","message":"Model o3-mini is enabled only for api versions 2024-12-01-preview and later"}}

It seems in the vscode dev tools it uses 2023 version though 2024 version is set as apiVersion

This also applies to o4-mini
However gpt-5, gpt-oss, gpt-4,1 seem to be working fine.

Solution
OpenAI2 = class extends BaseLLM -> 2024-12-01-preview

[Patrick-Erichsen]

I think your config.yaml is missing the env property: https://docs.continue.dev/customize/model-providers/top-level/azure#how-to-set-up-azure-ai-foundry-chat-models

Here is what I think yours should look like:

- name: GPT-o3-mini
  provider: azure
  model: o3-mini
  apiKey: #
  apiBase: https://#.cognitiveservices.azure.com/
  engine: o3-mini
  env:
    apiVersion: "2024-12-01-preview"

[LayerFort]

Is there any way I can force thinking mode for certain OpenAI (Azure) or Vertex models ?
Also being able to force grounding with google search on Gemini 2.5 Pro (Vertex) instead of using the search_web tool ?