mlsanityData to 14 Oct 2026

REFLEX with Jev for Efficient Selective Control in LLM Agents

arXiv 2609.26532 ↗First posted 22 Sep 2026cs.AIcited by 11

On the map: LLM agents and tool use → General LLM agent frameworks and benchmarks also fits: Reinforcement learning for LLM agents

Abstract (arXiv)

LLM agents often use generative models for bounded decisions, raising the question of when these decisions can be handled more efficiently without reducing task success. We study REFLEX, an agent architecture that uses Jev as a fast, typed decision layer and calls a strong LLM when confidence is low, or generation is required. On a frozen 100-task benchmark, REFLEX achieves 95% success with 72.7% fewer strong-model calls than a strong-only agent, with reductions persisting across three fallback families. Controlled interventions show that reliability depends on action-set size and near-valid alternatives near authorization boundaries. External BFCL and τ-style evaluations reveal limited advantages over a cheap generative cascade when ordinary routing is already highly accurate. These findings identify when selective control with Jev can reduce computation and where its benefits are limited.

Key facts. Cited by 11 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 7 of its 11 references that are in the map. First posted on arXiv 22 Sep 2026.

Builds on

7 of 11 references.

References · 7

  1. 2023FrugalGPT: How to Use Large Language Models While Reducing Cost and Improving Performancecited by 853
  2. 2024RouteLLM: Learning to Route LLMs with Preference Datacited by 682
  3. 2025τ2-Bench: Evaluating Conversational Agents in a Dual-Control Environmentcited by 627
  4. 2026LLMRouterBench: A Massive Benchmark and Unified Framework for LLM Routingcited by 38
  5. 2026The Confidence Dichotomy: Analyzing and Mitigating Miscalibration in Tool-Use Agentscited by 16
  6. 2025MICE for CATs: Model-Internal Confidence Estimation for Calibrating Agents with Toolscited by 15
  7. 2026Causal Evidence that Language Models use Confidence to Drive Behaviorcited by 10

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2026: 11.

11 papers.

  1. 2026Beyond Calibration: Do a Typed-Decision Model's Probabilities Obey the Probability Axioms?cited by 6
  2. 2026Calibrated Decision Models for Autonomous Penetration-Testing Harnesses: JEV and Laya as System One Decision Layers for LLM-Driven Pentest Agentscited by 4
  3. 2026Typed Decision Models: An Early Evidence Audit and Evaluation Checklistcited by 3
  4. 2026Benchmarking System One decision models against trained classifiers and language models for automated decision gatescited by 1
  5. 2026AnyJev Technical Reportcited by 1
  6. 2026Evaluating System One Models for Agent Security Decisions: Reliability, Calibration, and Selective Automationcited by 0
  7. 2026What Can Component-Replacement Evidence Establish? A Critical Scoping Review of Local Decisions in LLM Agentscited by 0
  8. 2026OmniMed-Jev: Calibrating LVLM Confidence for Trustworthy Medical Multimodal Decisions via System Onecited by 0
  9. 2026Readout Stability in Prefill-Only Decision Models:Zero-Label Prediction and Inference-Time Compute Allocationcited by 0
  10. 2026Can Jev be Your Q or Policy in Reinforcement Learning?cited by 0
  11. 2026Suggested Is Not Superseded: A Pre-Registered Evaluation of a System-One Model as a Memory-Retirement Gatecited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2026Knowing When to Ask for Help: Bayesian Self-Escalation in Hierarchical LLM Agents
  2. 2026A Control System, a Dataset, and a Recipe for Making Frozen LLM Agents Learn a Domain
  3. 2025Reflex First, Reflect Later: Latency-Aware Embodied LLM Agents for Dynamic Response
  4. 2026Multi-Head Latent Control: A Unified Interface for LLM Agent Decision Making
  5. 2026AutoGym: Blueprint-First Generation of Verifiable Agent Gyms
  6. 2026Fast Models, Slow Evidence: A Paired and Self-Audited Evaluation of System-1 Decision Models for LLM Agent Harnesses
  7. 2026Select-then-Solve: Paradigm Routing as Inference-Time Optimization for LLM Agents
  8. 2026Context Assembly as the Controlled Variable: A Control-Theoretic View of Harness Policies for Frozen LLM Agents