Same Patient, Different Order: Action-Level Reliability of Clinical LLM Agents Under Repeated Runs

• URL: http://arxiv.org/abs/2609.13582v1
• Date: Fri, 11 Sep 2026 22:44:35 GMT
• Title: Same Patient, Different Order: Action-Level Reliability of Clinical LLM Agents Under Repeated Runs
• Abstract summary: A clinical agent benchmark can report the same verdict on identical inputs while the agent files a materially different order on each run.<n>We introduce "same-input rerun", which replays a task with every input held fixed and compares the orders rather than the score.<n>The study establishes that action-level divergence exists and can pass unrecorded by the score, not that any rate generalises.
• Score: 6.442213381552777
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: A clinical agent benchmark can report the same verdict on identical inputs while the agent files a materially different order on each run. Such agents order tests, request medications and place referrals, yet benchmarks typically score one run per task and rarely ask whether identical inputs produce identical actions; MedAgentBench, the benchmark we use, scores a single attempt and says so. To measure this gap we introduce "same-input rerun", which replays a task with every input held fixed and compares the orders rather than the score, with six reliability metrics, and apply it to 1000 MedAgentBench runs across 50 tasks from its five write-capable families, two open-weight models below ten billion parameters quantised to four bits, and two temperatures. The study establishes that action-level divergence exists and can pass unrecorded by the score, not that any rate generalises. Under the 8B model at temperature 0.7, all 43 ordering groups emit a different set of orders across five identical runs, 26 emit the order on some runs and not others, and 28 record a different coded value, dose or analyte. In 22 of those 43 the benchmark reports the same failing verdict for materially different behaviour, as it does for all 10 divergent groups of the 4B model at 0.7. Orders also reach different endpoints across runs, one of which the record server rejects while the agent is told it succeeded. These findings motivate repeated-run evaluation, action-level stability reporting and execution-faithful environment feedback in clinical-agent benchmarks.

Related papers

• Closing the Consistency Gap: Self-Evolving Agents That Learn to Stay on Course [5.9756241643216255]
  Large language model (LLM)-powered agents can be accurate on average yet unreliable in production.<n>We present a self-evolving framework that reduces this gap by identifying unstable, low-consistency steps in agent trajectories.<n>On AppWorld with ReAct/GPT-4.1, our framework raises the fraction of tasks that succeed in all five runs by +16 points on same-task evaluation and +13 points on similar-task generalization.
  arXiv  Detail & Related papers  (2026-09-08T14:53:43Z)
• ExecRetrieval: Measuring the Functional-Correctness Gap in Code-Embedding Retrieval [0.0]
  Embedding-based code retrieval is a core component of coding agents and retrieval-augmented code generation.<n>Existing benchmarks do not plant controlled, execution-verified single-edit variants of each query's canonical implementation.<n>We introduce ExecRetrieval, 939 Python tasks each paired with one execution-verified canonical implementation and up to four execution-verified buggy distractors.
  arXiv  Detail & Related papers  (2026-09-01T20:55:17Z)
• Tencent WorkBuddy Bench: A Multi-Domain Coding-Agent Benchmark with Contamination-Resistant Task Construction [57.138342889101345]
  Tencent WorkBuddy Bench is a multi-domain evaluation suite for coding agents.<n>This report documents its construction methodology, scoring protocol, and a cross-model leaderboard.
  arXiv  Detail & Related papers  (2026-07-23T04:34:06Z)
• Fantastic Adaptive Taxonomies and How to Use Them [100.20614173157708]
  AdaMAST converts execution traces into a compact, evidence-grounded failure taxonomy.<n>No code is hand-authored, no trace human-annotated.<n>In agent-system search, taxonomy-coded diagnoses of failed candidates outperform free-form.
  arXiv  Detail & Related papers  (2026-07-17T17:41:16Z)
• PACE: A Proxy for Agentic Capability Evaluation [60.3743414796937]
  We introduce PACE, a framework that constructs proxy benchmarks by selecting instances from existing non-agentic evaluations.<n>Experiments across 14 models, 4 agentic benchmarks, and 19 non-agentic benchmarks show that PACE-Bench predicts agentic scores with leave-one-out cross-validation (LOOCV) mean absolute error (MAE) under 4%, Spearman correlation above 0.80, and pairwise model-ranking accuracy around 85%, all at much less than 1% of the full agentic evaluation cost.
  arXiv  Detail & Related papers  (2026-07-02T10:59:03Z)
• ClawArena-Team: Benchmarking Subagent Orchestration and Dynamic Workflows in Language-Model Agents [61.184954205350984]
  ClawArena-Team is a benchmark of 41 multi-turn, multimodal, multi-agent scenarios spanning 258 evaluation rounds and 72 staged updates.<n>All scoring is execution-based with no LLM judge: an overall score -- the Subagent-Management Score (SMS) -- multiplies task correctness by a least-privilege and modality-routing factor.
  arXiv  Detail & Related papers  (2026-06-30T06:06:08Z)
• SNARE: Adaptive Scenario Synthesis for Eliciting Overeager Behavior in Coding Agents [40.270213696031625]
  A coding agent executes a benign task as a sequence of shell, file, and network actions, any of which can quietly exceed the authorized scope while the task still completes.<n>We call this overeager behavior: the prompt is not adversarial and the run succeeds, yet an out-of-scope step can leak credentials or delete files.<n>Existing benchmarks miss it: task-completion suites credit any finished run, jailbreak suites probe adversarial prompts, and the one prior overeager benchmark applies a single fixed prompt to every agent-model pair.<n>We present SNARE, a pipeline that composes benign scenarios from reusable scope
  arXiv  Detail & Related papers  (2026-05-27T08:14:07Z)
• What Twelve LLM Agent Benchmark Papers Disclose About Themselves: A Pilot Audit and an Open Scoring Schema [0.0]
  We read twelve well-known LLM agent benchmark papers and recorded, dimension by dimension, what each paper actually says about how its evaluation was run.<n>We designed a small audit schema (five fields: benchmark identity, harness specification, inference settings, cost reporting, failure breakdown)<n>We score the disclosure of an agent run, not its correctness, and make no claim that disclosure implies a trustworthy result.
  arXiv  Detail & Related papers  (2026-05-20T17:02:36Z)
• ClawMark: A Living-World Benchmark for Multi-Turn, Multi-Day, Multimodal Coworker Agents [77.22389710754452]
  We introduce bench, a benchmark for coworker agents built around multi-turn multi-day tasks.<n>The current release contains 100 tasks across 13 professional scenarios, executed against five stateful sandboxed services.<n>The strongest model reaches 75.8 weighted score, but the best strict Task Success is only 20.0%, indicating that partial progress is common.
  arXiv  Detail & Related papers  (2026-04-26T16:05:02Z)
• When Agents Disagree With Themselves: Measuring Behavioral Consistency in LLM-Based Agents [0.0]
  ReAct-style agents produce 2.0--4.2 distinct action sequences per 10 runs on average, even with identical inputs.<n>Tasks with consistent behavior achieve 80--92% accuracy, while highly inconsistent tasks achieve only 25--60%.<n>Our results suggest that monitoring behavioral consistency during execution could enable early error detection and improve agent reliability.
  arXiv  Detail & Related papers  (2026-02-12T06:15:14Z)
• Towards a Science of Scaling Agent Systems [79.64446272302287]
  We formalize a definition for agent evaluation and characterize scaling laws as the interplay between agent quantity, coordination structure, modelic, and task properties.<n>We derive a predictive model using coordination metrics, that cross-validated R2=0, enabling prediction on unseen task domains.<n>We identify three effects: (1) a tool-coordination trade-off: under fixed computational budgets, tool-heavy tasks suffer disproportionately from multi-agent overhead, and (2) a capability saturation: coordination yields diminishing or negative returns once single-agent baselines exceed 45%.
  arXiv  Detail & Related papers  (2025-12-09T06:52:21Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.