Saved in:
Bibliographic Details
Main Authors: Sanchez, Gabby, Oommen, Sneha, Britto, Cassandra T., Wang, Di, Chiou, Jung-De, Spichkova, Maria
Format: Preprint
Published: 2026
Subjects:
Online Access:https://arxiv.org/abs/2604.01615
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866911562454794240
author Sanchez, Gabby
Oommen, Sneha
Britto, Cassandra T.
Wang, Di
Chiou, Jung-De
Spichkova, Maria
author_facet Sanchez, Gabby
Oommen, Sneha
Britto, Cassandra T.
Wang, Di
Chiou, Jung-De
Spichkova, Maria
contents This paper presents a systematic, cost-aware evaluation of large language models (LLMs) for receipt-item categorisation within a production-oriented classification framework. We compare four instruction-tuned models available through AWS Bedrock: Claude 3.7 Sonnet, Claude 4 Sonnet, Mixtral 8x7B Instruct, and Mistral 7B Instruct. The aim of the study was (1) to assess performance across accuracy, response stability, and token-level cost, and (2) to investigate what prompting methods, zero-shot or few-shot, are especially appropriate both in terms of accuracy and in terms of incurred costs. Results of our experiments demonstrated that Claude 3.7 Sonnet achieves the most favourable balance between classification accuracy and cost efficiency.
format Preprint
id arxiv_https___arxiv_org_abs_2604_01615
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Analysis of LLM Performance on AWS Bedrock: Receipt-item Categorisation Case Study
Sanchez, Gabby
Oommen, Sneha
Britto, Cassandra T.
Wang, Di
Chiou, Jung-De
Spichkova, Maria
Artificial Intelligence
Software Engineering
This paper presents a systematic, cost-aware evaluation of large language models (LLMs) for receipt-item categorisation within a production-oriented classification framework. We compare four instruction-tuned models available through AWS Bedrock: Claude 3.7 Sonnet, Claude 4 Sonnet, Mixtral 8x7B Instruct, and Mistral 7B Instruct. The aim of the study was (1) to assess performance across accuracy, response stability, and token-level cost, and (2) to investigate what prompting methods, zero-shot or few-shot, are especially appropriate both in terms of accuracy and in terms of incurred costs. Results of our experiments demonstrated that Claude 3.7 Sonnet achieves the most favourable balance between classification accuracy and cost efficiency.
title Analysis of LLM Performance on AWS Bedrock: Receipt-item Categorisation Case Study
topic Artificial Intelligence
Software Engineering
url https://arxiv.org/abs/2604.01615