Update README.md

README.md

@@ -30,6 +30,11 @@ language:
 
 When deploying "AlphaAI-Reason-V0" note that responses may contain <think> tokens, which serve as markers for the model's internal reasoning steps. If using the model in custom applications, these tokens may need to be preprocessed or filtered depending on your use case. Additionally, since the model is optimized for reasoning, responses may be more detailed compared to general-purpose language models.
 
+**Limitations**
+- Not a Large Model – Being a small 3B-parameter model, it may struggle with highly complex, multi-layered reasoning tasks that require deeper contextual awareness.
+- Performance Variability – While optimized, some edge cases requiring extensive memory, nuanced abstraction, or long-term dependencies may result in inconsistent outputs.
+- Processing Speed vs. Accuracy – Running the model on low-end devices may trade-off between response time and accuracy, depending on the quantization used.
+- Requires Post-Processing for <think> Tokens – The model explicitly demonstrates its reasoning, which may require custom handling when integrating into consumer-facing applications.
 
 **Ethical Considerations**