SupraBench: First benchmark for evaluating LLMs on supramolecular chemistry reasoning
Researchers introduce SupraBench, the first benchmark designed to systematically evaluate LLMs on supramolecular chemistry tasks including binding affinity prediction, top-binder selection, solvent identification, and host-guest description. The work also releases SupraPMC, a 16M-token corpus of supramolecular chemistry articles from Europe PMC to support domain adaptation. Evaluation of broad open and proprietary LLMs reveals substantial headroom across all tasks, with domain pretraining improving in-distribution regression but creating format compliance tradeoffs. The benchmark targets a narrow but practically important scientific domain where LLM acceleration could reduce days-long dry-lab verification cycles.
Related guides (1)
Related events (8)
MetaSyn benchmark reveals critical screening bottleneck in LLM-based meta-analysis pipelines
Researchers introduce MetaSyn, a dataset of 442 expert-curated meta-analyses from Nature Portfolio journals, paired with a 140k-article PubMed retrieval corpus, PI/ECO criteria, verified positives, and hard negatives. Benchmarking twelve pipeline configurations — nine RAG variants and a protocol-driven agent — shows that despite 90.9% retrieval recall at K=200, no system recovers more than 52.7% of ground-truth included studies. The core failure is LLMs' inability to reliably distinguish eligible studies from topically similar but criteria-failing distractors. The paper argues that end-to-end scores obscure where pipelines break down and proposes stage-attributed metrics.
Benchmarking study finds LLMs fail at counterintuitive probability problems despite strong standard performance
A new arXiv paper evaluates 8 state-of-the-art LLMs on discrete probability problems using two datasets: standard exercises (average accuracy 0.96) and counterintuitive exercises designed to trigger heuristic reasoning (average accuracy 0.59). The authors document token bias causing 20%+ performance drops when canonical problem formulations are disguised, and up to 34% degradation when misleading suggestions are embedded in prompts. The findings argue that current LLMs are not genuine probabilistic reasoners despite their success on advanced math benchmarks.
SoundnessBench: Benchmarking LLMs as Evaluators of ML Research Proposal Viability
SoundnessBench is a new benchmark of 1,099 machine-learning research proposals derived from ICLR submissions, labeled with reviewer soundness scores, designed to test whether LLMs can reliably distinguish methodologically sound research ideas from unsound ones. Evaluated across 12 frontier LLMs, the benchmark reveals a pervasive optimism bias: models systematically rate low-soundness proposals as sound under standard prompting, with aggressive prompting shifting errors from false positives to false negatives rather than eliminating them. Controls for data contamination, surface features, and human audit quality suggest the bias is not attributable to a single confounder. The authors conclude that current LLMs are not yet reliable as standalone first-gate evaluators of scientific rigor, a critical bottleneck for autonomous AI research agents.
FilBench: Benchmarking LLM Capabilities in Filipino Language
FilBench is a new benchmark introduced to evaluate large language models on their ability to understand and generate Filipino. The benchmark targets a historically underrepresented language in NLP evaluation suites, assessing both comprehension and generation tasks. This work addresses gaps in multilingual LLM evaluation coverage, particularly for Southeast Asian languages.
TxBench-PP: New benchmark reveals AI agents struggle with preclinical pharmacology decisions
Researchers introduce TxBench-PP (TherapeuticsBench Preclinical Pharmacology), a 100-evaluation benchmark testing AI agents on realistic small-molecule drug discovery tasks including mechanism-of-action reasoning, compound-target engagement, and translational efficacy. Agents receive real workflow snapshots and are graded deterministically on structured answers. Across 16 model-harness configurations and 4,800 trajectories, no system reliably succeeded; the best performer, Claude Opus 4.8 with the Pi harness, passed only 59.3% of endpoint attempts. The results suggest current frontier models are not yet deployment-ready for autonomous preclinical pharmacology decision-making.
Paper challenges LLM expert-level claims by measuring variance and error magnitude in code-based data analysis tasks
A new arXiv paper argues that standard LLM benchmarks overstate model capabilities by focusing on average performance on training-data-adjacent tasks while ignoring response variance and error magnitude. The authors introduce a novel benchmark requiring frontier LLMs to write code for data analysis tasks, comparing results against human expert submissions. Human experts outperformed the frontier LLM on average across multiple metrics and showed lower performance variability. The findings challenge the prevailing narrative that LLMs perform at human-expert level on knowledge economy tasks.
Phun-Bench: A Chinese benchmark for evaluating LLM phonological understanding
Researchers introduce Phun-Bench, a purpose-built benchmark for evaluating LLMs on phonological understanding in Chinese across three dimensions: Homophony, Rhyme, and Phonetic Similarity. The benchmark is designed to avoid rote-memorization shortcuts that plague existing phonological evals. Results show LLMs can recall correct pronunciations but fail to apply phonological knowledge flexibly as human speakers do, and the authors propose a hypothesis about the underlying mechanism of LLM phonological 'perception'.
Rethinking LLM Evaluation with 3C3H: AraGen Benchmark and Leaderboard
Hugging Face introduces AraGen, a new Arabic-language LLM benchmark and leaderboard built around the 3C3H evaluation framework (Correctness, Completeness, Conciseness, Helpfulness, Harmlessness, Honesty). The benchmark targets a gap in non-English LLM evaluation, specifically for Arabic, using a structured multi-criteria rubric rather than simple accuracy metrics. The leaderboard is hosted on Hugging Face and aims to provide a more holistic assessment of Arabic generative capabilities across frontier and open-weight models.