ARFBench (Text Input) - Start Time: 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 given as serialized text or through a time-series encoder. Accuracy on the Start Time questions (when the first anomaly begins), in percent; higher is better.
Metric: Accuracy (%) on the 56 Start Time questions (when the first anomaly begins, or no anomaly) of ARFBench, for the time series given as serialized text (the text-only LLMs) or through a time-series encoder (the time-series LLMs), with the series description 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 June 2027. 4 models tracked. Updated 7 Oct 2026.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Toto-1.0-Qwen3 32B (TSFM-LLM) | 35.7 | |
| 2 | GPT-5 | 29.6 | #178 |
| 3 | Qwen 3 32B | 27.3 | #972 |
| 4 | ChatTS 8B (TS-LLM) | 16.1 |
No result here: #1 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.
Interactive version: theaggregate.ai/benchmark?slug=arfbench-text-input-start-time · How It Works · Data refreshed daily, snapshot 2026-11-21.