ARFBench - Indicator: leaderboard

Time series question answering on software incident data in ARFBench (2026, Datadog): multiple-choice questions over multivariate production telemetry from real incidents, with the series shown as chart images. Accuracy on the Indicator questions (whether one anomaly leads or lags the other), in percent; higher is better.

Metric: Accuracy (%) on the 163 Indicator questions (whether one anomaly leads or lags the other) of ARFBench, for the time series rendered as chart images (Matplotlib plots, three images for paired questions), with the series description and channel names as text; few-shot multiple-choice prompting with shuffled answer choices, temperature 0.05 (default for models without the setting), reasoning models at medium effort with 2,000 output tokens, blank or invalid answers counted wrong; higher is better. Source: arxiv.org. Saturation forecast: Around February 2027. 11 models tracked. Updated 7 Oct 2026.

Top models

#ModelSettingScoreOverall rank
1GPT-561.3#178
2GPT-5.4Medium60.7#135
3Toto-1.0-QA-Experimental 32B (TSFM-VLM)60.1
4Gemini 3 Pro41.1#145
5Qwen 3 VL 32B (post-trained)39.3
6Qwen 3 VL 32B36.2#622
7Claude Opus 4.625.2#113
8GPT-4o23.9#751
9GPT-4.123.3#505
10Qwen 3 VL 8B17.8#917
11Claude Sonnet 4.517.2#261

No result here: #1 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.

Interactive version: theaggregate.ai/benchmark?slug=arfbench-indicator · How It Works · Data refreshed daily, snapshot 2026-11-21.