Error: qwen2.5-coder-7b-instruct-q4_k_m- 400

[sober060224]

Error Details

Model: qwen2.5-coder-7b-instruct-q4_k_m:latest
Provider: ollama
Status Code: 400

Error Output

"registry.ollama.ai/library/qwen2.5-coder-7b-instruct-q4_k_m:latest does not support tools"

Additional Context
Please add any additional context about the error here

[sestinj]

Related issues:

• Error: Qwen Coder 14B - 400 #9376 - Same "does not support tools" error with Ollama models
• "registry.ollama.ai/library/qwen2.5-coder:1.5b-base does not support tools" #5885 - Same error with qwen2.5-coder:1.5b-base (closed, fixed in docs: add ollama alternative to use system message tools #7718)
• deepseek-r1:8b does not support tools #5928 - Same error with deepseek-r1:8b (closed)

[RomneyDa]

Hello, could you share your config? Sounds like the error is on the API side and maybe we just need to tweak our tool support. Does the model support tools when run locally?

[sober060224]

Hi @RomneyDa, thanks for following up. Here's my config:

- name: Autodetect
  provider: ollama
  model: AUTODETECT
  apiBase: http://localhost:11434

Context: I manually created this model via a custom Modelfile that only contains FROM ./qwen2.5-coder-7b-instruct-q4_k_m.gguf without explicit tool-related parameters (e.g., PARAMETER tool_calling or Qwen's tool template). So it's likely the model doesn't support tools in my local setup.

Temporary workaround: After seeing sestinj's suggestion in related issues, I enabled "Only use system message tools" in Continue's settings. This lets me chat with the model again, though I'm not sure if full function-calling is working.

Question: Is there a reliable way to verify tool support for a locally imported GGUF model? (e.g., a specific API test or Modelfile checklist)

Thanks for your help!