update continue demo to use public HF models

[dtrawins]

🛠 Summary

https://openvino-doc.iotg.sclab.intel.com/hf/model-server/ovms_demos_code_completion_vsc.html

🧪 Checklist

• Unit tests added.
• The documentation updated.
• Change follows security best practices.
  ``

[Copilot]

Pull request overview

This PR updates the code_local_assistant demo to use publicly available Hugging Face OpenVINO model artifacts (pulled via --source_model) and refreshes the documentation/examples accordingly.

Changes:

• Bump OVMS Windows project version constant to 2026.1.0.
• Rewrite demos/code_local_assistant/README.md to reference OpenVINO-published HF models and updated run commands for Windows bare metal and Linux Docker.
• Add a new screenshot (vram.png) referenced from the demo README.

Reviewed changes

Copilot reviewed 2 out of 3 changed files in this pull request and generated 11 comments.

File	Description
windows_set_ovms_version.py	Updates the Windows OVMS project version constant to 2026.1.0.
demos/code_local_assistant/README.md	Reworks demo instructions to use OpenVINO public HF models and updates Continue configuration snippets.
demos/code_local_assistant/vram.png	Adds a new image used by the demo documentation.

[Copilot]

This NPU Docker example starts with mkdir c:\models (Windows path). In the Linux Docker section this should be a POSIX path (e.g., mkdir -p models) consistent with the $(pwd)/models bind mount.

[Copilot]

This Linux Docker command passes --model_repository_path c:\models, which is a Windows path and doesn’t match the container mount (/models). This will break model downloads/loading inside the container. Please use /models (or models if that’s the intended in-container path) consistently.

Suggested change

	mkdir c:\models
	docker run -d -p 8000:8000 --rm -e MOE_USE_MICRO_GEMM_PREFILL=0 --user $(id -u):$(id -g) -v $(pwd)/models:/models/:rw --device /dev/dri --group-add=$(stat -c "%g" /dev/dri/render* | head -n 1) -u $(id -u):$(id -g) \
	openvino/model_server:weekly \
	--model_repository_path c:\models --source_model OpenVINO/Qwen3-8B-int4-ov --task text_generation --target_device GPU --tool_parser hermes3 --reasoning_parser qwen3 --rest_port 8000 --model_name Qwen3-8B
	mkdir -p models
	docker run -d -p 8000:8000 --rm -e MOE_USE_MICRO_GEMM_PREFILL=0 --user $(id -u):$(id -g) -v $(pwd)/models:/models/:rw --device /dev/dri --group-add=$(stat -c "%g" /dev/dri/render* | head -n 1) -u $(id -u):$(id -g) \
	openvino/model_server:latest \
	--model_repository_path /models --source_model OpenVINO/Qwen3-8B-int4-ov --task text_generation --target_device GPU --tool_parser hermes3 --reasoning_parser qwen3 --rest_port 8000 --model_name Qwen3-8B

[Copilot]

The "Custom models" section uses python export_model.py ... but this README no longer includes any step to obtain export_model.py or install its requirements. As written, users will hit a missing-file/module error. Please add the download/install steps (or link to the canonical export_models instructions) before this example.

[Copilot]

This Linux Docker example starts with mkdir c:\models (Windows path), which will fail on Linux and doesn’t match the $(pwd)/models bind mount. Please update it to create the local models/ directory (e.g., mkdir -p models).

[Copilot]

This Windows section is labeled as a bash snippet but uses Windows CMD syntax/paths. Also, set MOE_USE_MICRO_GEMM_PREFILL=0 # ... is not valid in CMD (the # comment will be included in the value). Please switch the code fence to bat/PowerShell and format the set line/comment appropriately.

[Copilot]

The Linux Docker command mixes --user ... and -u ... (duplicated user flags) and uses openvino/model_server:latest while requesting --target_device GPU with /dev/dri mapped. For GPU instructions, other docs use the :latest-gpu image; please switch to the GPU image and remove the duplicate user option to avoid confusing or non-working copy/paste.

[Copilot]

In the Linux section this command starts with mkdir c:\models, which is a Windows path and will fail on Linux. It also doesn’t match the bind mount that uses $(pwd)/models. Please change this to creating the local models/ directory (e.g., mkdir -p models).

[rasapala]

Please verify my liked Copilot comments.

[dkalinowski]

Something is off here. Here in windows part, gpt doesnt require MOE_USE_MICRO_GEMM_PREFILL env, but below, in docker linux setup you pass MOE_USE_MICRO_GEMM_PREFILL to docker container. One of those is wrong

[dtrawins]

it is correct, only qwen3-moe models need this WA