mlsanityData to 14 Oct 2026

Should SLMs be in the Driver's Seat? A Study of Local Models as Interpreters in Cloud-Written Agent Programs

arXiv 2610.14209 ↗First posted 9 Oct 2026cs.AI cs.DCcited by 0

Not on the map. Why?

Abstract (arXiv)

The most accurate agents executing user-provided tasks are driven by large, cloud-hosted models, which solve them by reading the private data of the user. Agents powered by models small enough to run on edge devices keep that data local, but yield far worse results. A third option is to have cloud models write a program that devices execute: the data stays local, but as the program is written before any data is seen, it fails whenever a step depends on interpreting content that appears only at run time. We explore an intermediate architecture, in which the cloud-written program defers bounded questions over user data to a small local model through an ordinary function call. On AppWorld this improves the success rate of a code-only agent in all six combinations of backbone model and test partition we evaluate, by up to 14.9pp, recovering a substantial part of the gap to a cloud-hosted ReAct loop while keeping user data on the device.

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Builds on 40 of its 45 references that are in the map. First posted on arXiv 9 Oct 2026.

Builds on

40 of 45 references.

References · 40

  1. 2022ReAct: Synergizing Reasoning and Acting in Language Modelscited by 8,695
  2. 2020Scaling Laws for Neural Language Modelscited by 6,586
  3. 2023Toolformer: Language Models Can Teach Themselves to Use Toolscited by 4,182
  4. 2023HuggingGPT: Solving AI Tasks with ChatGPT and its Friends in Hugging Facecited by 1,190
  5. 2022Program of Thoughts Prompting: Disentangling Computation from Reasoning for Numerical Reasoning Taskscited by 1,076
  6. 2022PAL: Program-aided Language Modelscited by 652
  7. 2023ViperGPT: Visual Inference via Python Execution for Reasoningcited by 603
  8. 2024Executable Code Actions Elicit Better LLM Agentscited by 564
  9. 2022Visual Programming: Compositional visual reasoning without trainingcited by 516
  10. 2024AppWorld: A Controllable World of Apps and People for Benchmarking Interactive Coding Agentscited by 401
  11. 2023Describe, Explain, Plan and Select: Interactive Planning with Large Language Models Enables Open-World Multi-Task Agentscited by 398
  12. 2023Large Language Models as Tool Makerscited by 303
28 more
  1. 2025Defeating Prompt Injections by Designcited by 297
  2. 2025Small Language Models are the Future of Agentic AIcited by 270
  3. 2025Plan-and-Act: Improving Planning of Agents for Long-Horizon Taskscited by 195
  4. 2023ReWOO: Decoupling Reasoning from Observations for Efficient Augmented Language Modelscited by 184
  5. 2023An LLM Compiler for Parallel Function Callingcited by 179
  6. 2023ADaPT: As-Needed Decomposition and Planning with Language Modelscited by 177
  7. 2024TinyAgent: Function Calling at the Edgecited by 58
  8. 2024PAPILLON: Privacy Preservation from Internet-based and Local Language Model Ensemblescited by 57
  9. 2024Hammer: Robust Function-Calling for On-Device Language Models via Function Maskingcited by 55
  10. 2024Octopus v2: On-device language model for super agentcited by 47
  11. 2024AutoDroid-V2: Boosting SLM-based GUI Agents via Code Generationcited by 34
  12. 2024Octopus: On-device language model for function calling of software APIscited by 27
  13. 2025Agentic Plan Caching: Test-Time Memory for Fast and Cost-Efficient LLM Agentscited by 27
  14. 2025Minions: Cost-efficient Collaboration Between On-device and Cloud Language Modelscited by 20
  15. 2025Beyond ReAct: A Planner-Centric Framework for Complex Tool-Augmented LLM Reasoningcited by 17
  16. 2024CoAct: A Global-Local Hierarchy for Autonomous Agent Collaborationcited by 12
  17. 2024Octo-planner: On-device Language Model for Planner-Action Agentscited by 8
  18. 2025EcoAgent: An Efficient Device-Cloud Collaborative Multi-Agent Framework for Mobile Automationcited by 8
  19. 2025Federation of Agents: A Semantics-Aware Communication Fabric for Large-Scale Agentic AIcited by 7
  20. 2025TinyLLM: Evaluation and Optimization of Small Language Models for Agentic Tasks on Edge Devicescited by 6
  21. 2025AgentFlux: Decoupled Fine-Tuning & Inference for On-Device Agentic Systemscited by 3
  22. 2026PlanTwin: Privacy-Preserving Planning Abstractions for Cloud-Assisted LLM Agentscited by 3
  23. 2026PAAC: Privacy-Aware Agentic Device-Cloud Collaborationcited by 3
  24. 2025Advancing SLM Tool-Use Capability using Reinforcement Learningcited by 2
  25. 2026PseudoAct: Leveraging Pseudocode Synthesis for Flexible Planning and Action Control in Large Language Model Agentscited by 2
  26. 2026LLM-assisted Agentic Edge Intelligence Frameworkcited by 1
  27. 2026PlanCompiler: A Deterministic Compilation Architecture for Structured Multi-Step LLM Pipelinescited by 1
  28. 2026When Cloud Agents Meet Device Agents: Lessons from Hybrid Multi-Agent Systemscited by 1

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2026Nexus: An Execution Fabric for AI Agents Across Cloud, Edge, and Devices
  2. 2026Evaluating Local Language Model Agents for Reproducible Data Engineering: An Empirical Software Engineering Study of Mobility Workflows
  3. 2026Towards a Systems Foundation for Agentic Cloud Management
  4. 2026ECAS: An Edge-Controlled Agentic System for Validation-Gated Scientific Application Execution
  5. 2026SkillSmith: Enhancing Locally Deployed Agents via Automatic Skill Construction and Evolution
  6. 2024WorldCoder, a Model-Based LLM Agent: Building World Models by Writing Code and Interacting with the Environment
  7. 2026Harnessing LLMs as Agents: What Does It Cost?
  8. 2026Turn: A Language for Agentic Computation