Cannot get the thought process output using reasoning model with LM Studio backend

[iamWing]

I'm trying to setup the vscode extension using LMStudio as backend. While I can get the basic chat function working, I can't get the thought process block shows up in the chat panel.

Here is my config:

name: Local Agent
version: 1.0.0
schema: v1
models:
  - name: openai/gpt-oss-20b(local)
    provider: lmstudio
    model: openai/gpt-oss-20b
    apiBase: http://localhost:1234/v1/
    roles:
      - chat
      - edit
    capabilities:
      - tool_use
    defaultCompletionOptions:
      reasoning: true
      maxTokens: 131072
      contextLength: 131072
    requestOptions:
      extraBodyProperties:
        "max_completion_tokens": 131072
        "stream_options":
            "include_usage": true

  - name: qwen/qwen2.5-coder-14b(local)
    provider: lmstudio
    model: qwen/qwen2.5-coder-14b
    apiBase: http://localhost:1234/v1/
    roles:
      - autocomplete
    defaultCompletionOptions:
      maxTokens: 32768
      contextLength: 32768
    requestOptions:
      extraBodyProperties:
        "max_completion_tokens": 32768
        "stream_options":
            "include_usage": true

  - name: fastapply-1.5b-v1.0(local)
    provider: lmstudio
    model: fastapply-1.5b-v1.0
    apiBase: http://localhost:1234/v1/
    roles:
      - apply
    capabilities:
      - tool_use

  - name: qwen3-reranker-8b(local)
    provider: lmstudio
    model: qwen3-reranker-8b
    apiBase: http://localhost:1234/v1/
    roles:
      - rerank
    capabilities:
      - tool_use
    defaultCompletionOptions:
      maxTokens: 40960
      contextLength: 40960
    requestOptions:
      extraBodyProperties:
        "max_completion_tokens": 40960
        "stream_options":
            "include_usage": true

  - name: text-embedding-nomic-embed-text-v1.5(local)
    provider: lmstudio
    model: text-embedding-nomic-embed-text-v1.5
    apiBase: http://localhost:1234/v1/
    roles:
      - embed
    capabilities:
      - tool_use

My question is, did I do anything wrong on the config? Or is reasoning not supported when using LMStudio as backend?

Thanks!