bug(flaky): "Model not found" after user selects a model variant

[0xSage]

Cortex version

v75

Describe the Bug

This issue happens on Windows, the 1st time users try to pull each model. Meaning, it happens many times, each time user is trying to pull a new* model.

But, retrying the same request a second time works just fine.

(base) PS C:\Users\n\cortexcpp-nightly\models> cortex-nightly.exe pull llama3.1
Select an option
1. 8b-onnx
2. 8b-gguf
3. onnx
4. gguf

Select an option (1-4): 1
Model not found

After retrying:

(base) PS C:\Users\n\cortexcpp-nightly\models> cortex-nightly.exe pull llama3.1
Select an option
1. 8b-onnx
2. 8b-gguf
3. onnx
4. gguf

Select an option (1-4): 1
Validating download items, please wait..
...

Progresses as normal

Steps to Reproduce

No response

Screenshots / Logs

No response

What is your OS?

• [ ] MacOS
• [X] Windows
• [ ] Linux

What engine are you running?

• [ ] cortex.llamacpp (default)
• [ ] cortex.tensorrt-llm (Nvidia GPUs)
• [x] cortex.onnx (NPUs, DirectML)

[0xSage]

(base) PS C:\Users\n\cortexcpp-nightly\models\llama3.1> cortex-nightly.exe pull mistral
Select an option
1. 7b-tensorrt-llm-linux-ada
2. tensorrt-llm-linux-ada
3. 7b-tensorrt-llm-linux-ampere
4. tensorrt-llm-linux-ampere
5. 7b-tensorrt-llm-windows-ada
6. 7b-tensorrt-llm-windows-ampere
7. tensorrt-llm-windows-ampere
8. tensorrt-llm-windows-ada
9. onnx
10. gguf
11. 7b-onnx
12. 7b-gguf

Select an option (1-12): 10
Model not found