mlsanityData to 14 Oct 2026

BAGEN: Are LLM Agents Budget-Aware?

arXiv 2606.00198 ↗First posted 29 May 2026cs.LG cs.AI cs.CLcited by 13

On the map: LLM agents and tool use → General LLM agent frameworks and benchmarks

Abstract (arXiv)

While agents are increasingly spending more resources, today agent cost is mostly measured only after execution. A Budget-Aware Agent (BAGEN) should treat budget as an active control signal, rather than a passive cost metric. We first systematically define budget estimation as internal budgets (from agent computation) and external budgets (from agent actions). We then formalize budget-awareness as progressive interval estimation: at each step of a plan, an agent should predict an upper and lower bound on remaining budget, and alert when completion is unlikely. Scoring with a rollout-replay protocol, we find consistent failure patterns on four environments and five frontier agents: (1) strong agents do not necessarily have strong budget-awareness, with correlation r=0.35. (2) frontier models are consistently over-optimistic, continue spending on tasks that are unlikely to succeed, instead of alerting the user early. (3) budget-aware signal is actionable and trainable. Early stop saves 28-64% tokens on failed trajectories, and SFT+RL strengthens early stop and alert behavior. (4) precise interval calibration remains challenging, with interval coverage capping at 47% after SFT+RL. Project page: https://ragen-ai.github.io/bagen/

Key facts. Cited by 13 papers in the map (data through 14 Oct 2026). Placed in General LLM agent frameworks and benchmarks, one of 16 directions in LLM agents and tool use. Builds on 50 of its 57 references that are in the map. First posted on arXiv 29 May 2026.

Builds on

50 of 57 references.

Closest ancestors · 4

Same direction, newest first

  1. 2026Calibrate-Then-Act: Cost-Aware Exploration in LLM Agents18 Feb
  2. 2024Devil's Advocate: Anticipatory Reflection for LLM Agents25 May
  3. 2024Learning From Failure: Integrating Negative Examples when Fine-tuning Large Language Models as Agents18 Feb
  4. 2023AgentBench: Evaluating LLMs as Agents7 Aug

Other references · 46

  1. 2025Qwen3 Technical Reportcited by 8,820
  2. 2022ReAct: Synergizing Reasoning and Acting in Language Modelscited by 8,695
  3. 2023Reflexion: Language Agents with Verbal Reinforcement Learningcited by 4,276
  4. 2023Self-Refine: Iterative Refinement with Self-Feedbackcited by 3,442
  5. 2023SWE-bench: Can Language Models Resolve Real-World GitHub Issues?cited by 3,373
  6. 2023WebArena: A Realistic Web Environment for Building Autonomous Agentscited by 2,006
  7. 2024Scaling LLM Test-Time Compute Optimally can be More Effective than Scaling Model Parameterscited by 1,986
  8. 2022Language Models (Mostly) Know What They Knowcited by 1,856
  9. 2025Search-R1: Training LLMs to Reason and Leverage Search Engines with Reinforcement Learningcited by 1,444
  10. 2025s1: Simple test-time scalingcited by 1,341
  11. 2022WebShop: Towards Scalable Real-World Web Interaction with Grounded Language Agentscited by 1,251
  12. 2020ALFWorld: Aligning Text and Embodied Environments for Interactive Learningcited by 1,205
34 more
  1. 2017Constrained Policy Optimizationcited by 983
  2. 2023Can LLMs Express Their Uncertainty? An Empirical Evaluation of Confidence Elicitation in LLMscited by 974
  3. 2023SelfCheckGPT: Zero-Resource Black-Box Hallucination Detection for Generative Large Language Modelscited by 810
  4. 2023CRITIC: Large Language Models Can Self-Correct with Tool-Interactive Critiquingcited by 783
  5. 2016Distribution-Free Predictive Inference For Regressioncited by 668
  6. 2019Conformalized Quantile Regressioncited by 629
  7. 2017BranchyNet: Fast Inference via Early Exiting from Deep Neural Networkscited by 596
  8. 2016Adaptive Computation Time for Recurrent Neural Networkscited by 506
  9. 2018Accurate Uncertainties for Deep Learning Using Calibrated Regressioncited by 422
  10. 2018Reward Constrained Policy Optimizationcited by 417
  11. 2019Deep Evidential Regressioncited by 355
  12. 2017SkipNet: Learning Dynamic Routing in Convolutional Networkscited by 332
  13. 2019Self-Monitoring Navigation Agent via Auxiliary Progress Estimationcited by 176
  14. 2018High-Quality Prediction Intervals for Deep Learning: A Distribution-Free, Ensembled Approachcited by 117
  15. 2020Safe Reinforcement Learning in Constrained Markov Decision Processescited by 115
  16. 2019The Regretful Agent: Heuristic-Aided Navigation through Progress Estimationcited by 105
  17. 2024Large Language Models Must Be Taught to Know What They Don't Knowcited by 100
  18. 2019Evaluating and Calibrating Uncertainty Prediction in Regression Taskscited by 87
  19. 2024Can Large Language Models Faithfully Express Their Intrinsic Uncertainty in Words?cited by 87
  20. 2020Beyond Pinball Loss: Quantile Methods for Calibrated Uncertainty Quantificationcited by 78
  21. 2022Constrained Variational Policy Optimization for Safe Reinforcement Learningcited by 71
  22. 2026AgentRx: Diagnosing AI Agent Failures from Execution Trajectoriescited by 60
  23. 2020Constrained Upper Confidence Reinforcement Learningcited by 46
  24. 2024Large Language Models have Intrinsic Self-Correction Abilitycited by 46
  25. 2024Uncertainty of Thoughts: Uncertainty-Aware Planning Enhances Information Seeking in Large Language Modelscited by 39
  26. 2026How Do AI Agents Spend Your Money? Analyzing and Predicting Token Consumption in Agentic Coding Taskscited by 33
  27. 2020Learning to Stop While Learning to Predictcited by 31
  28. 2023Learning to Skip for Language Modelingcited by 19
  29. 2023Towards A Unified View of Answer Calibration for Multi-Step Reasoningcited by 8
  30. 2026Budget-Constrained Agentic Large Language Models: Intention-Based Planning for Costly Tool Usecited by 8
  31. 2026Adaptive Stopping for Multi-Turn LLM Reasoningcited by 7
  32. 2023Near-optimal Conservative Exploration in Reinforcement Learning under Episode-wise Constraintscited by 5
  33. 2024Safe Reinforcement Learning for Constrained Markov Decision Processes with Stochastic Stopping Timecited by 5
  34. 2026Quantifying the Accuracy and Cost Impact of Design Decisions in Budget-Constrained Agentic LLM Searchcited by 1

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2026: 13.

13 papers.

Sort:

By area:

  1. 2026CoffeeBench: Benchmarking Long-Horizon LLM Agents in Heterogeneous Multi-Agent Economiescited by 8
  2. 2026Fail-Fast, Restart-Smart: Early Failure Prediction and Restart for SWE Agentic Taskscited by 5
  3. 2026Doomed from the Start: Early Abort of LLM Agent Episodes via a Recall-Controlled Probe Cascadecited by 4
  4. 2026EarlyEval: Cheaper Agent Evaluation via Early Outcome Predictioncited by 3
  5. 2026AllocBench: Measuring Online Tool Allocation Capability in LLM Agentscited by 2
  6. 2026BAP-SQL: Budget-Aware Observation Planning for Agentic Text-to-SQLcited by 1
  7. 2026RECLAIM: Can Agents Reproduce the Claims of Machine Learning Papers?cited by 0
  8. 2026LLM Parkinsonism: Executive-Control Failure, Token-Inefficient Persistence, and an Uncertainty-Aware Global Executive Control Architecture for Autonomous Language-Model Agentscited by 0
  9. 2026A Competing-Hazards Systematization of Loss of Control in Autonomous Agentscited by 0
  10. 2026Range-GRPO: Policy Optimization via Pairwise Relations among Reward Intervalscited by 0
  11. 2026HERA: Harness-Environment Co-Evolution for Reliable Agentic Abstentioncited by 0
  12. 2026AgentTime: Can Agents Estimate and Control Their Own Runtime?cited by 0
1 more
  1. 2026On the Clock: Towards Punctual and Productive Time-Budgeted AI Agentscited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2026EcoAgent-Bench: Evaluating Economic Decision-Making in Budget-Constrained LLM Agents
  2. 2025Budget-Aware Tool Use Enables Effective Agent Scaling
  3. 2026ZEBRA: Zero-shot Budgeted Resource Allocation for LLM Orchestration
  4. 2026One Policy, Any Budget: Internalizing Budget-Aware Search via Reinforcement Learning
  5. 2026ContextBudget: Budget-Aware Context Management for Long-Horizon Search Agents
  6. 2026Spend Less, Reason Better: Budget-Aware Value Tree Search for LLM Agents
  7. 2025CostBench: Evaluating Multi-Turn Cost-Optimal Planning and Adaptation in Dynamic Environments for LLM Tool-Use Agents
  8. 2026STOCKTAKE: Measuring the Gap Between Perception and Action in LLM Agents with a Fair Oracle