In this paper, we compare the intent classification performance of several small-scale ( \(\sim \) 7B parameters range) open-source Large Language Models (LLMs), namely Mistral-7B-Instruct-v0.1, Mistral-7B-Instruct-v0.2 [12] and Meta-Llama-3-8B-Instruct [3] and fine-tune with a variety of training sample sizes (20, 100, 1k, 10k, and 50k samples) for multi-intent classification tasks using LoRA [11]. We also compare their results with two relatively smaller models, namely Phi-3-mini-4k-instruct ( \(\sim \) 3.8B parameters) [1] and bert-base-uncased ( \(\sim \) 110M parameters) [8], also fine-tuned on the same training chunks. We use the MultiWOZ 2.1 dataset [9], a widely used dataset for task-oriented dialogue systems that has 17 different intent classes. We get similar performance across these models while the training sample size is larger (50k samples) with Mistral-7B-Instruct-v0.2 having the best results with a micro-average f1-score of \(\sim \) 91% and macro-average f1-score of \(\sim \) 85%. This paper demonstrates that while using just 0.02% (100 training examples) of the fine-tuning data can achieve F-scores comparable to smaller models trained on much larger datasets, smaller models are more efficient in terms of training resources and inference time when 2% (or more) of the fine-tuning data (1,000 examples) is available.