mlsanityData to 7 Oct 2026

Task-Aware LLM Routing with Multi-Level Task-Profile-Guided Data Synthesis for Cold-Start Scenarios

arXiv 2604.09377 ↗First posted 10 Apr 2026cs.CLcited by 8

On the map: Synthetic data and self-improvement for LLMs → Code and instruction data synthesis

Abstract (arXiv)

Large language models (LLMs) exhibit substantial variability in performance and computational cost across tasks and queries, motivating routing systems that select models to meet user-specific cost-performance trade-offs. However, existing routers generalize poorly in cold-start scenarios where in-domain training data is unavailable. We address this limitation with a multi-level task-profile-guided data synthesis framework that constructs a hierarchical task taxonomy and produces diverse question-answer pairs to approximate the test-time query distribution. Building on this, we introduce TRouter, a task-type-aware router approach that models query-conditioned cost and performance via latent task-type variables, with prior regularization derived from the synthesized task taxonomy. This design enhances TRouter's routing utility under both cold-start and in-domain settings. Across multiple benchmarks, we show that our synthesis framework alleviates cold-start issues and that TRouter delivers effective LLM routing.

Key facts. Cited by 8 papers in the map (data through 7 Oct 2026). Placed in Code and instruction data synthesis, one of 13 directions in Synthetic data and self-improvement for LLMs. Builds on 27 of its 48 references that are in the map. First posted on arXiv 10 Apr 2026.

Builds on

27 of 48 references.

References · 27

  1. 2019DistilBERT, a distilled version of BERT: smaller, faster, cheaper and lightercited by 4,243
  2. 2016SQuAD: 100,000+ Questions for Machine Comprehension of Textcited by 4,115
  3. 2021Program Synthesis with Large Language Modelscited by 3,509
  4. 2024A Survey on LLM-as-a-Judgecited by 1,384
  5. 2023FrugalGPT: How to Use Large Language Models While Reducing Cost and Improving Performancecited by 791
  6. 2024RouteLLM: Learning to Route LLMs with Preference Datacited by 633
  7. 2023LLM-Blender: Ensembling Large Language Models with Pairwise Ranking and Generative Fusioncited by 550
  8. 2019Multi-News: a Large-Scale Multi-Document Summarization Dataset and Abstractive Hierarchical Modelcited by 431
  9. 2024Hybrid LLM: Cost-Efficient and Quality-Aware Query Routingcited by 354
  10. 2024Large Language Models for Data Annotation and Synthesis: A Surveycited by 201
  11. 2025Survey on Evaluation of LLM-based Agentscited by 190
  12. 2023Large Language Model Routing with Benchmark Datasetscited by 166
15 more
  1. 2024GraphRouter: A Graph-based Router for LLM Selectionscited by 154
  2. 2024A Note on Shumailov et al. (2024): `AI Models Collapse When Trained on Recursively Generated Data'cited by 114
  3. 2023Fly-Swat or Cannon? Cost-Effective Language Model Choice via Meta-Modelingcited by 94
  4. 2025Knowledge Graph Retrieval-Augmented Generation for LLM-based Recommendationcited by 56
  5. 2023EcoAssistant: Using LLM Assistant More Affordably and Accuratelycited by 47
  6. 2024Harnessing the Power of Multiple Minds: Lessons Learned from LLM Routingcited by 43
  7. 2024MetaLLM: A High-performant and Cost-efficient Dynamic Framework for Wrapping LLMscited by 39
  8. 2024Cost-Effective Online Multi-LLM Selection with Versatile Reward Modelscited by 36
  9. 2025OmniRouter: Budget and Performance Controllable Multi-LLM Routingcited by 35
  10. 2024Routoo: Learning to Route to Large Language Models Effectivelycited by 34
  11. 2025CARROT: A Cost Aware Rate Optimal Routercited by 34
  12. 2025Capability Instruction Tuning: A New Paradigm for Dynamic LLM Routingcited by 25
  13. 2025Anthropic Economic Index report: Uneven geographic and enterprise AI adoptioncited by 23
  14. 2024Enabling Efficient Batch Serving for LMaaS via Generation Length Predictioncited by 20
  15. 2025Enhancing Zero-Shot Image Recognition in Vision-Language Models through Human-like Concept Guidancecited by 1

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2026: 8.

8 papers.

  1. 2026NexForge: Scaling Agent Capabilities through Requirement-Driven Task Synthesis for LLMscited by 4
  2. 2026Uno-Orchestra: Parsimonious Agent Routing via Selective Delegationcited by 2
  3. 2026When Do LLMs Reason? A Dynamical Systems View via Entropy Phase Transitionscited by 2
  4. 2026HyDRA: Hybrid Dynamic Routing Architecture for Heterogeneous LLM Poolscited by 0
  5. 2026IR3DE: A Linear Router for Large Language Modelscited by 0
  6. 2026Self-Evolving Agents via Likelihood-Guided Tool-Space Optimizationcited by 0
  7. 2026Pretrain Once, Route Anywhere: Towards a Foundation Model for LLM Routingcited by 0
  8. 2026Routing Should Pay for Itself: Sparse Supervision for Economical LLM Routingcited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2025Rethinking Predictive Modeling for LLM Routing: When Simple kNN Beats Complex Learned Routers
  2. 2026Learning to Route LLMs from Implicit Cost-Performance Preferences via Meta-Learning
  3. 2026Dynamic Model Routing and Cascading for Efficient LLM Inference: A Survey
  4. 2025Adaptive LLM Routing under Budget Constraints
  5. 2026Scalable Prompt Routing via Fine-Grained Latent Task Discovery
  6. 2025When to Reason: Semantic Router for vLLM
  7. 2024TensorOpera Router: A Multi-Model Router for Efficient LLM Inference
  8. 2025Dynamic LLM Routing and Selection based on User Preferences: Balancing Performance, Cost, and Ethics