Can You Trust the Confidence? ConfBench for Vision-Language Models on Document Extraction
Priyashree Roy, Sujitha Martin, Mohammad Rostami, Spencer Romo, Renhao Xue, Bob Strahan, Diego A. Socolinsky, Boyi Xie, Md Mofijul Islam
cs.AI, cs.CL
Submitted: 2026-08-03
Code: https://github.com/awslabs/stickler
License: http://creativecommons.org/licenses/by-nc-sa/4.0/
The gist: Intelligent document processing (IDP) with vision-language models (VLMs) hinges on confidence scores trustworthy enough to route extractions between automation and human review.
Terminology
Abstract
Intelligent document processing (IDP) with vision-language models (VLMs) hinges on confidence scores trustworthy enough to route extractions between automation and human review. Existing document benchmarks are dominated by clean, high-quality samples, leaving low accuracy regions too sparse for calibration assessment. We introduce ConfBench, the first calibration-specific benchmark for key information extraction (KIE), built by applying 20 controlled degradation pipelines to a diverse document set, yielding 1,346 variants and 70K+ entity-level evaluations spanning the full accuracy spectrum. We evaluate four proprietary and three open-weight VLMs under verbalized and log-probability confidence estimation methods across three input modalities, and find: (i) OCR+Image modality results in more accurate confidence estimates; (ii) model capability is the dominant factor: within the Claude family confidence quality scales monotonically with capability, while across families parameter count is a poor predictor; (iii) calibration quality varies widely across models, from near-perfect to severely overconfident, and per-model post-hoc correction rescales these absolute confidence values for threshold-based routing without altering ranking-based operational metrics; and (iv) log-probability with first-token aggregation consistently outperforms mean-token and margin aggregations. We also introduce ECARB, a review-budget metric translating discriminative gains into operational savings. We release ConfBench publicly to enable systematic study of confidence estimators and calibration methods for trustworthy IDP application deployment.
Sources
- DocSplit: A Comprehensive Benchmark Dataset and Evaluation Approach for Document Packet Recognition and Splitting
- UNIKIE-BENCH: Benchmarking Large Multimodal Models for Key Information Extraction in Visual Documents
- DocILE Benchmark for Document Information Localization and Extraction
- RealKIE: Five Novel Datasets for Enterprise Key Information Extraction
- An Empirical Study Into What Matters for Calibrating Vision-Language Models
- Estimating LLM Uncertainty with Evidence
- A Survey of Calibration Process for Black-Box LLMs
- Object-Level Verbalized Confidence Calibration in Vision-Language Models via Semantic Perturbation
Related papers
- MAVEN-T: Reinforced Heterogeneous Distillation for Real-Time Multi-Agent Trajectory Prediction
- Model Discovery Agent: LLM-assisted Bayesian experiment design for data-efficient discovery of mechanistic world models
- Agents' Last Exam
- The Clinician's Veto: Navigating Trust, Liability, and Uncertainty in Autonomous AI Prescribing
- MindHelper: Closed-Loop Embodied Mental-State Reasoning for Precision Intervention
- Incumbent Advantage: Brand Bias and Cognitive Manipulation Dynamics in LLM Recommendation Systems