mlsanityData to 7 Oct 2026

Tool learning and function calling

456 papers7 this weekMedium confidence in this grouping: mostly one topic, with blurry edgesMedium confidence in this grouping: mostly one topic, with blurry edges · About

Keywords: tool, tools, llms, language, calling

Key facts. Tool learning and function calling: 456 papers on arXiv since 2022 (data through 7 Oct 2026). 252 of its 456 mapped papers (55%) cite “ToolLLM: Facilitating Large Language Models to Master 16000+ Real-world APIs” (2023, arXiv 2307.16789). 217 of its 456 mapped papers (48%) cite “Toolformer: Language Models Can Teach Themselves to Use Tools” (2023, arXiv 2302.04761). 34 new papers 17 Sep–7 Oct.

What’s new

17 Sep – 7 Oct 2026

34 new papers, 17 Sep – 7 Oct; 7 this week.

  1. This week · 7
  2. Loud Failures, Quiet Failures: Fault Detection and Recovery in Tool-Using Language Model Agents also fits: Software engineering coding agents
  3. How Do Agentic LLMs Decide to Call Tools? A Tool-Call Vector Shaped by Suppression
  4. Tool-calling retrieval versus vector RAG for a small Greek--English knowledge base: accuracy and robustness to how users type Greek also fits: Web, search, and research agents
  5. Quantization Effects on Tool-Failure Recovery Vary Across Prompts and Evaluation Designs
  6. Dynamic Budget Allocation for LLM Evaluation under Hard Resource Constraints
  7. SchemaFill: Efficient LLM Tool Calling via Slot-Parallel Speculative Decoding
  8. Turnslide: Scalable Multi-Turn Data Synthesis by Walking a Finite-State Machine
  9. Earlier · 27
  10. Topology-Consistent Task Planning over Cellular Workflow Complexes for LLM-based Agents also fits: General LLM agent frameworks and benchmarks
  11. Understanding and Mitigating Hallucination Escape in Tool-Using LLM Agents
  12. SHarP: Saliency-based Pruning of Agent Harnesses also fits: General LLM agent frameworks and benchmarks
All 34
  1. MLLMs Fail to Refuse when Using Tools Agentically
  2. Proxy Confidence: Auditing Black-Box LLM Agents with a Surrogate's Log-Probabilities also fits: Agent threats, guardrails and governance
  3. Toward SLM-based agentic task-tool intent matching also fits: General LLM agent frameworks and benchmarks
  4. What Does a Token Cost? A Mixture-of-Agents Measurement of Sufficient Per-Token Compute also fits: General LLM agent frameworks and benchmarks
  5. Choosing Before Acting: Comparative Value Estimation for Long-Horizon Tool-Use Agents
  6. Fast Models, Slow Evidence: A Paired and Self-Audited Evaluation of System-1 Decision Models for LLM Agent Harnesses also fits: General LLM agent frameworks and benchmarks
  7. Keyword Harnesses Fail Open: A Cheap Diagnostic Ladder for Tool-Use Claims in Small Language Models
  8. What Can Component-Replacement Evidence Establish? A Critical Scoping Review of Local Decisions in LLM Agents also fits: General LLM agent frameworks and benchmarks
  9. ContextRender: From Execution Dependencies to Agent Context also fits: Software engineering coding agents
  10. TokenCast: Forecasting Token Consumption During LLM Agent Execution
  11. One Readout, Many Repairs: Diffusion-Guided Hierarchical Search for Tool-Agent Repair also fits: Software engineering coding agents
  12. When Does Correction Become Repair? Mechanistic Auditing of Internal Interventions in Tool-Using LLMs
  13. Action-Space Shaping for LLM Agents: Measuring and Mitigating Tool-Schema Bias
  14. PDEU-Bench: Benchmarking the Personalized Planning Lifecycle of Tool-Calling LLM Agents
  15. Quantization Thresholds Replicate, Failure Modes Do Not: A Three-Model Study of Agentic Tool Use in Polish from 8-bit to 2-bit
  16. ToolSearcher: Optimizing Tool Selection at Scale via Reinforcement Learning also fits: Reinforcement learning for LLM agents
  17. Learning from Failures: Heterogeneous Graph Memory for Small Language Model Tool-Using Agents also fits: Agent memory systems
  18. TwinCheck: Evidence-Grounded Negative-Twin Verification for Stateful Tool Agents also fits: Software engineering coding agents
  19. Measuring the Serving Stack Instead of the Model: Hidden Confounds in Local Tool-Use Evaluation
  20. MCP-GRANITE Benchmark: GRANularity Interface TEsting for MCP-Based LLM Agents
  21. Lookahead-R: Budget-Aware Tool Retrieval via Execution-Centric Planning
  22. Constrained Decoding Eliminates Structural Failures in Small LLMs but Reveals a Scale-Dependent Semantic Gap
  23. What Stops a Small Language Model From Driving a Database Agent also fits: General LLM agent frameworks and benchmarks
  24. MATCH: Model-Aware Tool Learning with Curriculum Scheduling and Hierarchically Gated Rewards

Foundations

Most cited by papers here

  1. ToolLLM: Facilitating Large Language Models to Master 16000+ Real-world APIs2023cited by 252 here
  2. Toolformer: Language Models Can Teach Themselves to Use Tools2023cited by 217 here
  3. Gorilla: Large Language Model Connected with Massive APIs2023cited by 213 here
  4. API-Bank: A Comprehensive Benchmark for Tool-Augmented LLMs2023cited by 111 here
  5. ToolAlpaca: Generalized Tool Learning for Language Models with 3000 Simulated Cases2023cited by 105 here

Lineage

The papers this direction builds on, oldest first.

Each paper on the line cites the one before.

  1. 2023Toolformer: Language Models Can Teach Themselves to Use Toolscited by 217 here
  2. 2023Gorilla: Large Language Model Connected with Massive APIscited by 213 here
  3. 2023ToolAlpaca: Generalized Tool Learning for Language Models with 3000 Simulated Casescited by 105 here
  4. 2023ToolLLM: Facilitating Large Language Models to Master 16000+ Real-world APIscited by 252 here
  5. 2023MetaTool Benchmark for Large Language Models: Deciding Whether to Use Tools and Which to Usecited by 65 here
  6. 2024StableToolBench: Towards Stable Large-Scale Benchmarking on Tool Learning of Large Language Modelscited by 34 here
  7. 2024APIGen: Automated Pipeline for Generating Verifiable and Diverse Function-Calling Datasetscited by 52 here
  8. 2024ToolACE: Winning the Points of LLM Function Callingcited by 46 here
  9. 2025ToolRL: Reward is All Tool Learning Needscited by 36 here
15 more it builds on, not in this chain
  1. 2023Reflexion: Language Agents with Verbal Reinforcement Learningcited by 42 here
  2. 2023API-Bank: A Comprehensive Benchmark for Tool-Augmented LLMscited by 111 here
  3. 2023Tool Learning with Foundation Modelscited by 36 here
  4. 2023ToolkenGPT: Augmenting Frozen Language Models with Massive Tools via Tool Embeddingscited by 58 here
  5. 2023GPT4Tools: Teaching Large Language Model to Use Tools via Self-instructioncited by 36 here
  6. 2023ToolQA: A Dataset for LLM Question Answering with External Toolscited by 45 here
  7. 2023WebArena: A Realistic Web Environment for Building Autonomous Agentscited by 36 here
  8. 2023AgentBench: Evaluating LLMs as Agentscited by 34 here
  9. 2024EASYTOOL: Enhancing LLM-based Agents with Concise Tool Instructioncited by 31 here
  10. 2024AnyTool: Self-Reflective, Hierarchical Agents for Large-Scale API Callscited by 38 here
  11. 2024Seal-Tools: Self-Instruct Tool Learning Dataset for Agent Tuning and Detailed Benchmarkcited by 33 here
  12. 2024Tool Learning with Large Language Models: A Surveycited by 35 here
  13. 2024τ-bench: A Benchmark for Tool-Agent-User Interaction in Real-World Domainscited by 94 here
  14. 2024xLAM: A Family of Large Action Models to Empower AI Agent Systemscited by 32 here
  15. 2025APIGen-MT: Agentic Pipeline for Multi-Turn Data Generation via Simulated Agent-Human Interplaycited by 29 here

The 24 most cited of 1,054.

Browse

30 of 456, ranked 6–35

  1. 6MetaTool Benchmark for Large Language Models: Deciding Whether to Use Tools and Which to Use202365
  2. 7ToolkenGPT: Augmenting Frozen Language Models with Massive Tools via Tool Embeddings202358
  3. 8APIGen: Automated Pipeline for Generating Verifiable and Diverse Function-Calling Datasets202452
  4. 9ToolACE: Winning the Points of LLM Function Calling202446
  5. 10ToolQA: A Dataset for LLM Question Answering with External Tools202345
  6. 11AnyTool: Self-Reflective, Hierarchical Agents for Large-Scale API Calls202438
  7. 12Tool Learning with Foundation Models202336
  8. 13GPT4Tools: Teaching Large Language Model to Use Tools via Self-instruction202336
  9. 14Tool Learning with Large Language Models: A Survey202435
  10. 15StableToolBench: Towards Stable Large-Scale Benchmarking on Tool Learning of Large Language Models202434
All 30
  1. 16Seal-Tools: Self-Instruct Tool Learning Dataset for Agent Tuning and Detailed Benchmark202433
  2. 17EASYTOOL: Enhancing LLM-based Agents with Concise Tool Instruction202431
  3. 18TALM: Tool Augmented Language Models202228
  4. 19Large Language Models as Tool Makers202325
  5. 20NESTFUL: A Benchmark for Evaluating LLMs on Nested Sequences of API Calls202424
  6. 21Granite-Function Calling Model: Introducing Function Calling Abilities via Multi-task Learning of Granular Tasks202422
  7. 22ART: Automatic multi-step reasoning and tool-use for large language models202322
  8. 23ComplexFuncBench: Exploring Multi-Step and Constrained Function Calling under Long-Context Scenario202521
  9. 24MCP-Universe: Benchmarking Large Language Models with Real-World Model Context Protocol Servers202521
  10. 25API-BLEND: A Comprehensive Corpora for Training and Benchmarking API LLMs202421
  11. 26MINT: Evaluating LLMs in Multi-turn Interaction with Tools and Language Feedback202319
  12. 27LiveMCPBench: Can Agents Navigate an Ocean of MCP Tools?202518
  13. 28MCP-Bench: Benchmarking Tool-Using LLM Agents with Complex Real-World Tasks via MCP Servers202518
  14. 29Hammer: Robust Function-Calling for On-Device Language Models via Function Masking202418
  15. 30From Exploration to Mastery: Enabling LLMs to Master Tools via Self-Driven Interactions202418
  16. 31RoTBench: A Multi-Level Benchmark for Evaluating the Robustness of Large Language Models in Tool Learning202417
  17. 32Making Language Models Better Tool Learners with Execution Feedback202317
  18. 33MCP-RADAR: A Multi-Dimensional Benchmark for Evaluating Tool Use Capabilities in Large Language Models202515
  19. 34SMART: Self-Aware Agent for Tool Overuse Mitigation202513
  20. 35Learning to Use Tools via Cooperative and Interactive Agents202413