ARFBench: Benchmarking Time Series Question Answering Ability for Software Incident Response

TL;DR

ARFBench is a comprehensive benchmark for evaluating the ability of foundation models to perform time series question answering in software incident response, highlighting the performance gap and potential for hybrid approaches.

Contribution

The paper introduces ARFBench, a new large-scale TSQA benchmark with a novel hybrid model prototype and analysis of model-human complementarity.

Findings

Frontier VLMs outperform existing baselines.

GPT-5 achieves 62.7% accuracy and 51.9% F1.

Hybrid TSFM + VLM models match frontier model performance.

Abstract

Time series question-answering (TSQA), in which we ask natural language questions to infer and reason about properties of time series, is a promising yet underexplored capability of foundation models. In this work, we present ARFBench, a TSQA benchmark that evaluates the understanding of multimodal foundation models (FMs) on time series anomalies prevalent in software incident data. ARFBench consists of 750 questions across 142 time series and 5.38M data points from 63 production incidents sourced exclusively from internal telemetry at Datadog. We evaluate leading proprietary and open-source LLMs, VLMs, and time series FMs and observe that frontier VLMs perform markedly better than existing baselines; the leading model (GPT-5) achieves a 62.7% accuracy and 51.9% F1. We next demonstrate the promise of specialized multimodal approaches. We develop a novel TSFM + VLM hybrid prototype which we post-train on a small set of synthetic and real data that yields comparable overall F1 and accuracy with frontier models. Lastly, we find models and human domain experts exhibit complementary strengths. We define a model-expert oracle, a best-of-2 oracle selector over model and expert answers, yielding 82.8% F1 and 87.2% accuracy and establishing a new superhuman frontier for future TSQA models. The benchmark is available at https://huggingface.co/datasets/Datadog/ARFBench.