Description of onnxRuntimeGenAI.QNN Long-Term Inference Output Anomaly Issue

Bug Description and Reproduction

Describe the Bug

After long-term inference using onnxRuntimeGenAI.QNN, nonsensical output or repeated punctuation may occur.
EnableCaching has been disabled during initialization:

var options = new OnnxRuntimeGenAIChatClientOptions  
{  
    StopSequences = Array.Empty<string>(),  
    PromptFormatter = TestPromptFormatter,  
    EnableCaching = false  
};  

To Reproduce

Steps to reproduce the issue:

1. Initialize the model
2. Repeat the inference process (Start inference → Wait for inference to end → Start inference)

Expected behavior

Normal inference output.

Desktop (please complete the following information)

• OS: Windows 11 Home 25H2 26200.6588
• OnnxRuntimeGenAI.QNN: 0.10.0
• NPU Driver: 30.0.140.1000/30.0.145.1000

Additional Context

Add any other relevant information about the issue here.

Start by reproducing the repeated inference cycle described in the issue with EnableCaching disabled on Windows 11, using OnnxRuntimeGenAI.QNN 0.10.0 and the listed NPU drivers. Compare output after repeated Start inference → Wait for inference to end cycles; done means normal output remains consistent rather than becoming nonsensical or repeating punctuation.