# AllocBench: Measuring Online Tool Allocation Capability in LLM Agents

arXiv: https://arxiv.org/abs/2607.23332 · HTML: https://mlsanity.com/paper/2607.23332/

## Key facts

- Cited by 2 papers in the map (data through 14 Oct 2026).
- Placed in Tool learning and function calling, one of 16 directions in LLM agents and tool use.
- Builds on 26 of its 33 references that are in the map.
- First posted on arXiv 25 Jul 2026.

## Placed in

- LLM agents and tool use / [Tool learning and function calling](https://mlsanity.com/agents/tool-learning-and-function-calling/)

## Builds on (most cited first, 26 of 26)

- [Proximal Policy Optimization Algorithms](https://mlsanity.com/paper/1707.06347/) (2017)
- [QLoRA: Efficient Finetuning of Quantized LLMs](https://mlsanity.com/paper/2305.14314/) (2023)
- [Kimi k1.5: Scaling Reinforcement Learning with LLMs](https://mlsanity.com/paper/2501.12599/) (2025)
- [Large Language Models as Tool Makers](https://mlsanity.com/paper/2305.17126/) (2023)
- [SWE-RL: Advancing LLM Reasoning via Reinforcement Learning on Open Software Evolution](https://mlsanity.com/paper/2502.18449/) (2025)
- [Reasoning or Reciting? Exploring the Capabilities and Limitations of Language Models Through Counterfactual Tasks](https://mlsanity.com/paper/2307.02477/) (2023)
- [Asymmetric Actor Critic for Image-Based Robot Learning](https://mlsanity.com/paper/1710.06542/) (2017)
- [Tool Learning with Large Language Models: A Survey](https://mlsanity.com/paper/2405.17935/) (2024)
- [Competitive Programming with Large Reasoning Models](https://mlsanity.com/paper/2502.06807/) (2025)
- [CREATOR: Tool Creation for Disentangling Abstract and Concrete Reasoning of Large Language Models](https://mlsanity.com/paper/2305.14318/) (2023)
- [Planning, Creation, Usage: Benchmarking LLMs for Comprehensive Tool Utilization in Real-World Complex Scenarios](https://mlsanity.com/paper/2401.17167/) (2024)
- [TroVE: Inducing Verifiable and Efficient Toolboxes for Solving Programmatic Tasks](https://mlsanity.com/paper/2401.12869/) (2024)
- [LILO: Learning Interpretable Libraries by Compressing and Documenting Code](https://mlsanity.com/paper/2310.19791/) (2023)
- [Budget-Aware Tool Use Enables Effective Agent Scaling](https://mlsanity.com/paper/2511.17006/) (2025)
- [ReGAL: Refactoring Programs to Discover Generalizable Abstractions](https://mlsanity.com/paper/2401.16467/) (2024)
- [Do LLM Agents Have Regret? A Case Study in Online Learning and Games](https://mlsanity.com/paper/2403.16843/) (2024)
- [Calibrate-Then-Act: Cost-Aware Exploration in LLM Agents](https://mlsanity.com/paper/2602.16699/) (2026)
- [Benchmarking LLM Tool-Use in the Wild](https://mlsanity.com/paper/2604.06185/) (2026)
- [BAGEN: Are LLM Agents Budget-Aware?](https://mlsanity.com/paper/2606.00198/) (2026)
- [Spend Less, Reason Better: Budget-Aware Value Tree Search for LLM Agents](https://mlsanity.com/paper/2603.12634/) (2026)
- [When Agents Commit Too Soon: Diagnosing Premature Commitment in LLM Agents](https://mlsanity.com/paper/2606.22936/) (2026)
- [Budget-Constrained Agentic Large Language Models: Intention-Based Planning for Costly Tool Use](https://mlsanity.com/paper/2602.11541/) (2026)
- [Library Learning Doesn't: The Curious Case of the Single-Use "Library"](https://mlsanity.com/paper/2410.20274/) (2024)
- [Utility-Guided Agent Orchestration for Efficient LLM Tool Use](https://mlsanity.com/paper/2603.19896/) (2026)
- [When Greedy Wins: Emergent Exploitation Bias in Meta-Bandit LLM Training](https://mlsanity.com/paper/2509.24923/) (2025)
- [On Time, Within Budget: Constraint-Driven Online Resource Allocation for Agentic Workflows](https://mlsanity.com/paper/2605.06110/) (2026)

## Built on by (most cited first, 2 of 2)

- [When Stale Constraints Go Unchecked: Budgeted Verification Failures in Inherited Agent Memory](https://mlsanity.com/paper/2608.25553/) (2026)
- [Towards Reliable AI Data Scientists: Data Agents with Workflow Harnesses](https://mlsanity.com/paper/2609.35255/) (2026)

## Abstract (arXiv)

Creating a reusable tool is an investment: an agent pays a fixed cost now in exchange for the potential of future reuse. Therefore, a user should prefer an agent that creates a small number of highly reusable tools, rather than many one-offs. We introduce a paired benchmark that tests whether LLM agents exhibit conscious allocation behavior under a fixed budget in two contexts: an abstract text-based formulation and a code-construction task. We find that every frontier model we test---Claude Haiku, Claude Opus, GPT-5.4-mini, and GPT-5.6 Sol---acts near-optimally in the abstract framing but fails to transfer this ability to script-writing. Through further experiments, we identify the particular failure modes for each model. Notably, the first three models fail even when the scripts are not evaluated, while GPT-5.6 Sol stays selective under that weaker manipulation and collapses only at full construction. Furthermore, an open-source Qwen model policy-trained for abstract allocation generalizes this ability across held-out lexical variations, but sees no improvement at script allocation. Together, these results establish online tool allocation as a significant capability boundary, even for modern frontier models.