Model fails to call tools, returns tool call in text instead

Before submitting your bug report

• I've tried finding an answer on the Continue docs site
• I'm not able to find an open issue that reports the same bug
• I've seen the troubleshooting guide on the Continue Docs

Relevant environment info

- OS: Linux / NixOS
- Continue version: 2.0.0
- IDE version:
- Model: gemma4 e4b
- config:
  
name: Local Assistant
version: 1.0.0
schema: v1
models:
  - name: Gemma 4
    provider: openai
    model: OpenVINO/gemma-4-E4B-it-int4-ov
    apiKey: unused
    apiBase: http://localhost:8000/v3
    roles:
      - chat
      - edit
      - apply
    capabilities:
      - tool_use
      - image_input
    requestOptions:
      extraBodyProperties:
        chat_template_kwargs:
          enable_thinking: false
    chatOptions:
      baseAgentSystemMessage: You are a useful AI Agent and give concise answers without any repetition

context:
  - provider: code
  - provider: docs
  - provider: diff
  - provider: terminal
  - provider: problems
  - provider: folder
  - provider: codebase
  
  OR link to agent in Continue hub:

Description

Hi everyone. I am trying to make use of tool calling for the model. As far as I am aware gemma4 and my Backend wich is OpenVINO run on my local machine inside a docker container support tool calling. Say I make a query like "Search the web for foo", then I get

So I guess the model tries to call the search_web tool but somehow it fails silently. I have not found any similar reports. Somebody has an idea where to look first?

To reproduce

No response

Log output

The payload names no source files or tests. Start by reproducing the request with the shown Continue configuration, the OpenVINO-compatible apiBase, and the tool_use capability; compare the model's textual response with the expected search_web tool execution. Done means the reported setup invokes the tool instead of returning its call as text.