These small parameter models are not as effective as GPT or other powerful models. They can experience issues such as hallucinations, sentences being cut off, and other inconsistencies.