mlsanityData to 7 Oct 2026

TS-Haystack: A Multi-Task Retrieval Benchmark for Long-Context Time-Series Reasoning

arXiv 2602.14200 ↗First posted 15 Feb 2026cs.LGcited by 1

On the map: LLM reasoning → Time series reasoning

Abstract (arXiv)

Time Series Language Models (TSLMs) promise reasoning over real-world temporal data, but their ability to retrieve and reason over long time-series remains largely untested. We introduce TS-Haystack, a multi-domain retrieval benchmark with ten event-grounded question-answering tasks over contexts from 100 seconds to 24 hours, spanning direct retrieval, temporal reasoning, multi-step reasoning, and contextual anomaly detection. Existing TSLMs exhibit severe long-context degradation: accuracy declines with context length, direct-tokenization models run out of memory beyond 100 seconds on high-rate signals, and time-interval-grounded tasks collapse toward near-zero accuracy when increasing the time-series lengths, aligning with existing literature on text and multi-modal long context retrieval. An agentic retrieval framework using specialized time-series classifier tools matches or outperforms SoTA TSLMs on 9 of 10 tasks, highlighting agentic retrieval as a promising approach for long-context TSLMs.

Key facts. Cited by 1 paper in the map (data through 7 Oct 2026). Placed in Time series reasoning, one of 17 directions in LLM reasoning. Builds on 38 of its 58 references that are in the map. First posted on arXiv 15 Feb 2026.

Builds on

38 of 58 references.

Closest ancestors · 2

Same direction, newest first

  1. 2025OpenTSLM: Time-Series Language Models for Reasoning over Multivariate Medical Text- and Time-Series Data2 Oct
  2. 2024ChatTS: Aligning Time Series with LLMs via Synthetic Data for Enhanced Understanding and Reasoning4 Dec

Other references · 36

  1. 2017Attention Is All You Needcited by 59,932
  2. 2024The Llama 3 Herd of Modelscited by 14,192
  3. 2022ReAct: Synergizing Reasoning and Acting in Language Modelscited by 8,355
  4. 2023Visual Instruction Tuningcited by 7,792
  5. 2016XGBoost: A Scalable Tree Boosting Systemcited by 5,120
  6. 2023Mamba: Linear-Time Sequence Modeling with Selective State Spacescited by 4,616
  7. 2022Flamingo: a Visual Language Model for Few-Shot Learningcited by 4,337
  8. 2023Lost in the Middle: How Language Models Use Long Contextscited by 3,383
  9. 2023MMMU: A Massive Multi-discipline Multimodal Understanding and Reasoning Benchmark for Expert AGIcited by 2,044
  10. 2022A Time Series is Worth 64 Words: Long-term Forecasting with Transformerscited by 1,981
  11. 2019Towards VQA Models That Can Readcited by 1,696
  12. 2021Perceiver: General Perception with Iterative Attentioncited by 1,089
24 more
  1. 2025OpenAI GPT-5 System Cardcited by 910
  2. 2018The UCR Time Series Archivecited by 421
  3. 2018The UEA multivariate time series classification archive, 2018cited by 330
  4. 2024VideoAgent: Long-form Video Understanding with Large Language Model as Agentcited by 323
  5. 2024Titans: Learning to Memorize at Test Timecited by 303
  6. 2024BABILong: Testing the Limits of LLMs with Long Context Reasoning-in-a-Haystackcited by 211
  7. 2024GIFT-Eval: A Benchmark For General Time Series Forecasting Model Evaluationcited by 188
  8. 2025Chronos-2: From Univariate to Universal Forecastingcited by 158
  9. 2023Walking Down the Memory Maze: Beyond Context Limit through Interactive Readingcited by 129
  10. 2020Deep Learning for Time Series Forecasting: Tutorial and Literature Surveycited by 117
  11. 2024Large Language Models for Time Series: A Surveycited by 113
  12. 2025Recursive Language Modelscited by 94
  13. 2024ChatTime: A Unified Multimodal Time Series Foundation Model Bridging Numerical and Textual Datacited by 92
  14. 2024MileBench: Benchmarking MLLMs in Long Contextcited by 86
  15. 2022Self-supervised Learning for Human Activity Recognition Using 700,000 Person-days of Wearable Datacited by 71
  16. 2024Needle In A Multimodal Haystackcited by 56
  17. 2024TSFM-Bench: A Comprehensive and Unified Benchmark of Foundation Models for Time Series Forecastingcited by 55
  18. 2024Multimodal Needle in a Haystack: Benchmarking Long-Context Capability of Multimodal Large Language Modelscited by 53
  19. 2024TimeSeriesExam: A time series understanding examcited by 47
  20. 2021HiRID-ICU-Benchmark -- A Comprehensive Machine Learning Benchmark on High-resolution ICU Datacited by 38
  21. 2025ITFormer: Bridging Time Series and Natural Language for Multi-Modal QA with Large-Scale Multitask Datasetcited by 35
  22. 2025AudioMarathon: A Comprehensive Benchmark for Long-Context Audio Understanding and Efficiency in Audio LLMscited by 19
  23. 2024CAPTURE-24: A large dataset of wrist-worn activity tracker data collected in the wild for human activity recognitioncited by 18
  24. 2022Classification of ECG based on Hybrid Features using CNNs for Wearable Applicationscited by 2

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2026: 1.

1 paper.

  1. 2026TimeRLM: Recursive Language Models Enable Precise Anomaly Localization in Long-Context Time-Seriescited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2026TS-Skill: A Benchmark for Evaluating Analytical Skills in Time-Series Question Answering
  2. 2025TS-Agent: Understanding and Reasoning Over Raw Time Series via Iterative Insight Gathering
  3. 2025Time-MQA: Time Series Multi-Task Question Answering with Context Enhancement
  4. 2025Haystack Engineering: Context Engineering for Heterogeneous and Agentic Long-Context Evaluation
  5. 2026Thoth: Mid-Training Bridges LLMs to Time Series Understanding
  6. 2026From Recognition to Understanding: Unlocking Cognitive Time Series Reasoning with LLMs
  7. 2026TimeSage-MT: A Multi-Turn Benchmark for Evaluating Agentic Time Series Reasoning
  8. 2026Rationale-Grounded In-Context Learning for Time Series Reasoning with Multimodal Large Language Models