LLM-as-Judge papers, 2023–2026

Key findings

• LLM-as-Judge went from 8 papers in 2023 to 219 in 2026, roughly 27x in three years.
• NeurIPS publishes the most LLM-as-Judge work: 173 papers from 2023 to 2026, 43% of the four-venue total.
• The fastest-growing overlap is with AI Agents: papers tagged with both rose from 1 in 2023 to 78 in 2026.
• Among the 5 topics under Evaluation and Benchmarks, it ranks #2 by 2026 paper count; the fastest grower in the group, LLM Benchmarks, is up 154x versus 27x here.

Representative papers

Spread across years and venues so every edition shows up

• NeurIPS 2026A Free Lunch in LLM Compression: Revisiting Retraining after Pruning
• ICML 2026A Judge-Aware Ranking Framework for Evaluating Large Language Models without Ground Truth
• ICLR 2026Addressing Pitfalls in the Evaluation of Uncertainty Estimation Methods for Natural Language Generation
• CVPR 2026Pixels Dont Lie But Your Detector Might Bootstrapping MLLM-as-a-Judge for
• NeurIPS 2025$\texttt{BetaConform}$: Efficient MAP Estimation of LLM Ensemble Judgment Performance with Prior Transfer
• ICML 2025Active Evaluation Acquisition for Efficient LLM Benchmarking
• ICLR 2025An Empirical Analysis of Uncertainty in Large Language Model Evaluations
• CVPR 2025ASHiTA: Automatic Scene-grounded HIerarchical Task Analysis
• NeurIPS 2024BertaQA: How Much Do Language Models Know About Local Culture?
• ICML 2024CHAI: Clustered Head Attention for Efficient LLM Inference
• ICLR 2024PandaLM: An Automatic Evaluation Benchmark for LLM Instruction Tuning Optimization
• NeurIPS 2023Judging LLM-as-a-Judge with MT-Bench and Chatbot Arena
• ICML 2023SAM operates far from home: eigenvalue regularization as a dynamical phenomenon
• ICLR 2023On the complexity of nonsmooth automatic differentiation
• NeurIPS 2026A Stratified Multi-Rater Evaluation of LLM-Based Virtual Standardized Patients with a Deployed Data-Generation Platform
• ICML 2026Are LLM Evaluators Really Narcissists? Sanity Checking Self-Preference Evaluations
• ICLR 2026AutoMetrics: Approximate Human Judgments with Automatically Generated Evaluators
• CVPR 20263DrawAgent Teaching LLM to Draw in 3D with Early Contrastive
• NeurIPS 2025Adaptive Prediction-Powered AutoEval with Reliability and Efficiency Guarantees
• ICML 2025Evaluating Judges as Evaluators: The JETTS Benchmark of LLM-as-Judges as Test-Time Scaling Evaluators

All LLM-as-Judge papers

219 papers, page 3 of 5

101. ICML 2026LLM Watermark Evasion via Bias Inversion
102. ICML 2026LLM-based Embeddings: Attention Values Encode Sentence Semantics Better Than Hidden States
103. ICML 2026Margin-Adaptive Confidence Ranking for Reliable LLM Judgement
104. ICML 2026Mitigating Perceptual Judgment Bias in Multimodal LLM-as-a-Judge via Perceptual Perturbation and Reward Modeling
105. ICML 2026Mitigating Translationese Bias in Multilingual LLM-as-a-Judge via Disentangled Information Bottleneck
106. ICML 2026No Free Lunch: Non-Asymptotic Analysis of Prediction-Powered Inference
107. ICML 2026Nonparametric LLM Evaluation from Preference Data
108. ICML 2026On the Limits of LLM Adaptability: Impact of Model-Internalized Priors on Annotation Task Performance
109. ICML 2026Order Matters: Unveiling the Hidden Impact of Macro Placement Sequences via Proxy-Guided LLM Evolution
110. ICML 2026PPDL: LLM-Based Flows as Probabilistic Programs
111. ICML 2026PPT-Eval: A Benchmark for Computer-Use Agents on PowerPoint Tasks
112. ICML 2026Prediction-Powered Risk Monitoring of Deployed Models for Detecting Harmful Distribution Shifts
113. ICML 2026Pressure Reveals Character: Behavioural Alignment Evaluation at Depth
114. ICML 2026ProbeLLM: Automating Principled Diagnosis of LLM Failures
115. ICML 2026ProtocolBench: Which LLM MultiAgent Protocol to Choose?
116. ICML 2026Quantifying Biases in LLM-as-a-Judge Evaluations
117. ICML 2026Quartet II: Accurate LLM Pre-Training in NVFP4 by Improved Unbiased Gradient Estimation
118. ICML 2026REAL: Regression-Aware Reinforcement Learning for LLM-as-a-Judge
119. ICML 2026Reasoning Is Not Free: Robust Adaptive Cost-Efficient Routing for LLM-as-a-Judge
120. ICML 2026Reward Modeling from Natural Language Human Feedback
121. ICML 2026RLCracker: Evaluating the Worst-Case Vulnerability of LLM Watermarks with Adaptive RL Attacks
122. ICML 2026SCOPE: Selective Conformal Optimized Pairwise LLM Judging
123. ICML 2026Security--Fidelity Tradeoffs: The Hidden Cost of Prompt Injection Defense
124. ICML 2026SelfJudge: Faster Speculative Decoding via Self-Supervised Judge Verification
125. ICML 2026SpreadsheetArena: Decomposing Preference in LLM Generation of Spreadsheet Workbooks
126. ICML 2026Talk, Judge, Cooperate: Gossip-Driven Indirect Reciprocity in Self-Interested LLM Agents
127. ICML 2026Training-Free Hashing-Based Attention via Binary Principal Components
128. ICML 2026Turning Bias into Bugs: Bandit-Guided Style Manipulation Attacks on LLM Judges
129. ICML 2026UniRRM: Unified Reasoning Reward Models Across Languages and Evaluation Paradigms
130. ICML 2026Who can we trust? LLM-as-a-jury for Comparative Assessment
131. ICML 2026Zero-source LLM Hallucination Detection with Human-like Criteria Probing
132. NeurIPS 20263D-PLOT-LLM: Part-Level Object Tokens for 3D Large Language Models
133. NeurIPS 2026A Free Lunch in LLM Compression: Revisiting Retraining after Pruning
134. NeurIPS 2026A Stratified Multi-Rater Evaluation of LLM-Based Virtual Standardized Patients with a Deployed Data-Generation Platform
135. NeurIPS 2026Accelerating LLM Pre-Training through Flat-Direction Dynamics Enhancement
136. NeurIPS 2026ADAPT: Hybrid Prompt Optimization for LLM Feature Visualization
137. NeurIPS 2026AdvJudge-Zero: Binary Decision Flips in LLM-as-a-Judge via Adversarial Control Tokens
138. NeurIPS 2026AgentAbstain: Do LLM Agents Know When Not to Act?
139. NeurIPS 2026Alignment Is Not Enough for Safe Medical LLM Evaluation
140. NeurIPS 2026Are LLM Safety Judges Policy-Invariant? A Three-Principle Stress-Test
141. NeurIPS 2026Ask in the Crowd: Differentially Private LLM Inference via Dummy-Augmented Shuffling
142. NeurIPS 2026ASPI: Seeking Ambiguity Clarification Amplifies Prompt Injection Vulnerability in LLM Agents
143. NeurIPS 2026Auditing the Judge: Human-Grounded Bias Discovery, Quantification, and Mitigation in LLM Judges
144. NeurIPS 2026Autoregressive Diffusion World Models for Off-Policy Evaluation of LLM Agents
145. NeurIPS 2026BayesJudge: Uncertainty-Aware Bayesian Meta-Evaluation of Human and LLM Judgments
146. NeurIPS 2026Benchmarks Are Not Atomic: Composition-Aware LLM Evaluation using BenchHub
147. NeurIPS 2026Block-OBS-GS: Exact Per-Block Joint Brain Surgery with Gauss–Seidel Refinement for LLM Pruning
148. NeurIPS 2026BMAttn: Block-Aligned Mixed-Precision Attention Quantization for LLM Inference
149. NeurIPS 2026Can an LLM Reason Like a Lawyer? Benchmarking the ability of LLMs to map the facts of a case to the elements of the applicable legal rule
150. NeurIPS 2026CapTrack: Multifaceted Evaluation of Forgetting in LLM Post-Training

FAQ

How many LLM-as-Judge papers were published at NeurIPS, ICLR, ICML and CVPR from 2023 to 2026?

401 in total: 8 in 2023, 48 in 2024, 126 in 2025, 219 in 2026.

How many LLM-as-Judge papers are at NeurIPS 2026?

NeurIPS 2026 has 88 LLM-as-Judge papers, versus 53 the year before.

How many LLM-as-Judge papers are at ICLR 2026?

ICLR 2026 has 62 LLM-as-Judge papers, versus 37 the year before.

How many LLM-as-Judge papers are at ICML 2026?

ICML 2026 has 64 LLM-as-Judge papers, versus 33 the year before.

How many LLM-as-Judge papers are at CVPR 2026?

CVPR 2026 has 5 LLM-as-Judge papers, versus 3 the year before.

Which conference publishes the most LLM-as-Judge papers?

NeurIPS, with 173 papers from 2023 to 2026.