Feature Request: Support Qwen3 Coder 30B Local Model for Tool Calls

[desugar-64]

Before submitting your bug report

• I've tried using the "Ask AI" feature on the Continue docs site to see if the docs have an answer
• I believe this is a bug. I'll try to join the Continue Discord for questions
• 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: macOS
- Continue version:1.1.71
- IDE version: VS Code 1.102.3 (Universal)
- Model: qwen/qwen3-coder-30b
- config:
  
name: Local Assistant
version: 1.0.0
schema: v1
models:
  - name: LM Studio
    provider: lmstudio
    apiBase: http://192.168.0.45:6970/v1
    capabilities:
      - tool_use
    model: qwen/qwen3-coder-30b
    roles:
      - chat
      - edit
      - apply
  - name: Nomic Embed Text
    provider: lmstudio
    model: nomic-ai/nomic-embed-text-v1.5-GGUF
    roles:
      - embed
context:
  - provider: code
  - provider: docs
  - provider: diff
  - provider: terminal
  - provider: problems
  - provider: folder
  - provider: codebase

  
  OR link to assistant in Continue hub:

Description

I’m trying to use the new Qwen3 Coder 30B model locally with Continue.dev via LM Studio. While the model loads and responds, it seems that tool call requests (e.g., run_terminal_command, grep_search) are not handled, the tool call structure is printed out literally instead of being executed.

It looks like the plugin doesn’t recognize or properly handle tool call outputs from this model. Would be great to add support for Qwen3 Coder 30B as it performs well and supports function/tool calling.

Let me know if any debug logs or configs would help. Thanks for the awesome project!

To reproduce

No response

Log output

[desugar-64]

https://x.com/lmstudiodevs/status/1951101732269773021?s=46

Looks like it might be a model issue. The LM Studio released a fix for this. I’ll take a look.

[Pizzaschachtel]

That beta version (0.3.21-beta) of lmstudio doesn't solve the problem in my case :( It still exists

    provider: lmstudio
    model: lmstudio-community/Qwen3-Coder-30B-A3B-Instruct-MLX-4bit
    roles:
      - chat
      - edit
      - apply
      - summarize
    capabilities:
      - tool_use
    defaultCompletionOptions:
      contextLength: 40960

Model: qwen3-coder-30b-a3b-instruct-mlx-4bit

[desugar-64]

Yeah, it didn’t help for me either.
LM Studio 0.3.21 (Build 2)

[Pizzaschachtel]

I also want to mention, that the bug occured on intellij for Mac. It is not just a bug on vscode.

Edit:
Could the reason for the error perhaps be the fact that qwen3-coder uses XML instead of JSON for tool calls? Qwen3-coder provides its own parser for this purpose. The question is whether continue.dev should implement this parser, or whether this should be handled elsewhere.
See here:
https://huggingface.co/Qwen/Qwen3-Coder-30B-A3B-Instruct/blob/main/qwen3coder_tool_parser.py

[XiXinping]

Hey there! I'd like to add that I've encountered a similar error when using the VSCode extension on Linux and trying to connect to an Ollama instance running Qwen3-Coder.

OS: Manjaro Linux
Continue Version: v1.1.71
Model: qwen3-coder:30b
VSCode Version: 1.101.2

Here's my config.yaml:

name: Local Assistant
version: 1.0.0
schema: v1
models:
  - name: Qwen3-Coder
    provider: ollama
    model: qwen3-coder:30b
    apiBase: http://192.168.1.84:11434
    roles:
      - chat
      - autocomplete
      - edit
      - apply
  - name: Qwen2.5-Coder
    provider: ollama
    model: qwen2.5-coder:1.5b
    apiBase: http://192.168.1.84:11434
    roles:
      - chat
      - autocomplete
      - edit
      - apply
  - name: Nomic Embed
    provider: ollama
    model: nomic-embed-text:latest
    apiBase: http://192.168.1.84:11434
    roles:
      - embed
context:
  - provider: code
  - provider: docs
  - provider: diff
  - provider: terminal
  - provider: problems
  - provider: folder
  - provider: codebase

[Pizzaschachtel]

Hey there! I'd like to add that I've encountered a similar error when using the VSCode extension on Linux and trying to connect to an Ollama instance running Qwen3-Coder.

OS: Manjaro Linux Continue Version: v1.1.71 Model: qwen3-coder:30b VSCode Version: 1.101.2

Here's my config.yaml:

name: Local Assistant
version: 1.0.0
schema: v1
models:
  - name: Qwen3-Coder
    provider: ollama
    model: qwen3-coder:30b
    apiBase: http://192.168.1.84:11434
    roles:
      - chat
      - autocomplete
      - edit
      - apply
  - name: Qwen2.5-Coder
    provider: ollama
    model: qwen2.5-coder:1.5b
    apiBase: http://192.168.1.84:11434
    roles:
      - chat
      - autocomplete
      - edit
      - apply
  - name: Nomic Embed
    provider: ollama
    model: nomic-embed-text:latest
    apiBase: http://192.168.1.84:11434
    roles:
      - embed
context:
  - provider: code
  - provider: docs
  - provider: diff
  - provider: terminal
  - provider: problems
  - provider: folder
  - provider: codebase

Hi @XiXinping in this case you need to add

capabilities:
      - tool_use

as show in my example #6913 (comment)
However, you’ll likely run into the same issue that the author and I experienced.

[XiXinping]

So this is using Ollama, but I was able to get tool-calling working with qwen3-coder by editing the model template to use qwen3's template which does support tool-calling.

This is my Modelfile:

FROM qwen3-coder:30b

TEMPLATE """


{{- $lastUserIdx := -1 -}}
{{- range $idx, $msg := .Messages -}}
{{- if eq $msg.Role "user" }}{{ $lastUserIdx = $idx }}{{ end -}}
{{- end }}
{{- if or .System .Tools }}<|im_start|>system
{{ if .System }}
{{ .System }}
{{- end }}
{{- if .Tools }}

# Tools

You may call one or more functions to assist with the user query.

You are provided with function signatures within <tools></tools> XML tags:
<tools>
{{- range .Tools }}
{"type": "function", "function": {{ .Function }}}
{{- end }}
</tools>

For each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:
<tool_call>
{"name": <function-name>, "arguments": <args-json-object>}
</tool_call>
{{- end -}}
<|im_end|>
{{ end }}
{{- range $i, $_ := .Messages }}
{{- $last := eq (len (slice $.Messages $i)) 1 -}}
{{- if eq .Role "user" }}<|im_start|>user
{{ .Content }}<|im_end|>
{{ else if eq .Role "assistant" }}<|im_start|>assistant
{{ if (and $.IsThinkSet (and .Thinking (or $last (gt $i $lastUserIdx)))) -}}
<think>{{ .Thinking }}</think>
{{ end -}}
{{ if .Content }}{{ .Content }}
{{- else if .ToolCalls }}<tool_call>
{{ range .ToolCalls }}{"name": "{{ .Function.Name }}", "arguments": {{ .Function.Arguments }}}
{{ end }}</tool_call>
{{- end }}{{ if not $last }}<|im_end|>
{{ end }}
{{- else if eq .Role "tool" }}<|im_start|>user
<tool_response>
{{ .Content }}
</tool_response><|im_end|>
{{ end }}
{{- if and (ne .Role "assistant") $last }}<|im_start|>assistant
{{ end }}
{{- end }}
"""

PARAMETER stop "User:"
PARAMETER stop "Assistant:"

I then ran:
ollama create qwen3-coder-tool:30b -f Modelfile

I then edited my config.yaml accordingly and it seemed to work.