Search AI news

Measuring how AI agents handle everyday wet-lab science

OpenAI introduces BioBasicsBench, a 100-task benchmark measuring how well AI agents interpret raw experimental data and handle everyday wet-lab science.

ProcessPrefBench: Evaluating Cross-Task Transfer of Individual Process Preferences from Naturalistic Human-AI Interaction

arXiv:2611.06065v1 Announce Type: new Abstract: In natural human-AI dialogue, users' corrections, requests, and self-descriptions carry reusable signals about how they want work to be done. These signals reveal individual process preferences, user-specific standards for evidence, depth, analytical focus, procedure, and delivery. We ask whether a preference revealed in one dialogue can shape an assistant's later answer when the task does not restate it. ProcessPrefBench uses three months of natural dialogues from two knowledge workers to build 66 source-target cases separately judged for source evidence and target applicability. Sources are preselected, so the benchmark tests application rather than retrieval. We test three models with target-only, full-history, and distilled-preference prompts. Target-specific 0-2 rubrics score preference realization, and a task gate flags severe failures. Compared with target-only prompting, full-history and imperative distilled-preference prompting raise preference realization by 0.204 and 0.244 points on the 0-2 scale, respectively, with no reliable advantage from distillation. Fabricated-evidence failures rise from 5.8% under target-only prompting to 10.1% under full-history and 10.4% under imperative prompting. ProcessPrefBench highlights a tension in learning from dialogue, where drawing on prior interactions can improve preference realization yet also invite fabricated evidence. We hope it helps build assistants that honor a user's process preferences without sacrificing factual reliability.

A Controlled Experimental Evaluation of Configuration and Cache Architecture in LLM Serving Engines

arXiv:2611.05991v1 Announce Type: new Abstract: Multi-agent LLM applications repeat most of their prompt tokens across requests, and serving engines reuse the cached state of shared prefixes. vLLM and SGLang reuse it at different granularities, and benchmarks of their default deployments report large performance gaps, usually credited to the cache. Such comparisons rarely control for other configuration differences. One of them, whether a long prompt is prefilled in chunks, is set by each engine's defaults from the hardware and the release, and it confounds the comparison. We built a reproducible testbed on Kubernetes and a seeded multi-agent workload generator, and compared the two engines under vendor defaults and with chunked prefill matched. SGLang's configuration is the same in both arms, which makes it an internal control across independently provisioned deployments. Under defaults, vLLM's median time to first token is up to 64x SGLang's. Matching one setting removes 78% of the capacity gap where that gap is largest and 30 to 67% of the latency gap. Near saturation the remaining latency gap reflects queue growth. A fluid queue model driven only by a 4.4% throughput shortfall reproduces it, and shows that such a gap scales with the length of the benchmark run. What survives matching is a tail-latency advantage of 2.4 to 6.4x at equal throughput, consistent with cache-aware ordering of waiting requests. We derive six recommendations for comparing serving engines.

Say Less, Mean Better: Training-Free Verbosity Control for Language Models

arXiv:2611.04107v1 Announce Type: new Abstract: Large language models (LLMs) often generate responses longer than necessary, increasing inference latency and computational cost. While prompting can encourage conciseness, controlling output length without compromising response completeness remains challenging. Aggressive early stopping can reduce generation costs but risks terminating responses prematurely. We introduce MinimalLM, a training-free framework for reliable verbosity control that combines uncertainty-aware, length-dependent stop-token adjustment with structural completeness guardrails. MinimalLM separates the decision of when to stop from whether stopping is structurally appropriate, coordinating response-level instructions, decoding-time intervention, and completion checks without modifying model parameters or requiring auxiliary models. We evaluate MinimalLM across three open-source LLMs and 330 questions, producing 5,940 responses across baseline and ablation configurations. MinimalLM reduces generated tokens by 47.6%, accelerates individual responses by 19-68%, and eliminates structurally truncated outputs in the primary evaluation. Despite substantial token savings, it achieves an average quality score of 8.04/10, compared with 8.42/10 for unmodified models and 6.46/10 for length-matched truncation. Evaluations across diverse subject domains (such as healthcare, programming, law, and science) demonstrate robust efficiency gains, while ablations establish the complementary roles of decoding control and completeness constraints. Our findings highlight that efficient generation requires more than producing fewer tokens: it requires controlling when generation ends while preserving response integrity. MinimalLM offers a lightweight, training-free approach toward this goal.

Stress-Testing the Robustness of LLM-Based Causal Inference

arXiv:2611.03656v1 Announce Type: new Abstract: Why do some of the world's most powerful Artificial Intelligence (AI) systems still struggle to reason about cause and effect? Causal discovery, the ability to discern the underlying drivers of a system, is fundamental to scientific inquiry and has traditionally relied on structured data and statistical methods. Recently, Large Language Models (LLMs) have been explored as an alternative paradigm for causal inference, leveraging their broad knowledge and reasoning capabilities. Yet, despite their linguistic fluency, LLMs often rely on surface-level pattern recognition rather than genuine causal logic. While it is known that LLMs stumble in this domain, we lack a precise map of where, why, and how significantly these failures occur. In this paper, we propose a systematic stress-test framework across six critical dimensions: model scale, causal graph complexity, prompting strategy, metadata quality, data quality, and sensitivity to uncertainty. Our findings reveal a significant "optimism bias" in LLMs; models consistently overestimate causal links, leading to low precision across the board. Furthermore, we demonstrate that traditional evaluation metrics can be deceptive: while reasoning-heavy models lead in performance, the highest scores often occur when uncertainty in inference is overlooked. Once uncertainty is integrated into the evaluation, performance drops, providing a more honest reflection of causal reliability. We also find that, while scaling model size improves outcomes, these gains are fragile; performance degrades sharply as the quality of metadata decreases. This work serves as a practical guide to the blind spots of LLMs, offering a clear-eyed assessment of when these tools can be integrated into a causal discovery pipeline and when they are likely to lead researchers astray.

Exposure Is Not Commitment: Double-Stratified Evaluation of Latent Monitors for Tool-Using Agents

arXiv:2610.37679v1 Announce Type: new Abstract: Tool-using language agents are vulnerable to indirect prompt injection, and a growing literature probes model internals to catch unsafe actions before they execute, with sharply conflicting results. We argue much of the conflict is an evaluation artifact, and propose a double-stratification protocol: separate injection exposure from behavioral commitment, in both the evaluation and the training population. On the primary Qwen AgentDojo grid, monitors trained on the naive mixed unsafe-versus-rest contrast mostly learn exposure: they flag exposed trajectories whether or not the agent complies, and barely predict which cases turn unsafe. Retrained on exposed cases only, the same probe families predict the model's own unsafe tool call up to two s

Nvidia releases all-in-one adaptable Nemotron 3.5 audio transcription and translation model

Nvidia Corp. today released a leaderboard-winning open artificial intelligence model for multilingual audio transcription and translation, along with an open data pipeline for fine-tuning Nemotron 3.5 Transcribe. Traditional transcription AI models follow a workflow that detects speech, transcribes it, identifies and labels different speakers, then answers questions about the transcription. Transcribe does all of this […] The post Nvidia releases all-in-one adaptable Nemotron 3.5 audio transcription and translation model appeared first on SiliconANGLE.

Fix evaluation bias to unlock better models on Amazon Bedrock

Property Finder found that 96.8% of their vision-AI evaluation labels were inherited from the incumbent model, hiding its true accuracy. See how they built unbiased ground truth, applied evolutionary prompt optimization, and selected a more accurate, cost-effective model on Amazon Bedrock, cutting cost by 86% across 11 image-quality attributes.

Exowatt beats frontier models on accuracy with SageMaker serverless model customization

In this post, we show how Exowatt used Amazon SageMaker AI serverless model customization to build a repeatable AI pipeline that produces task-specific fine-tuned models for two deployment targets, edge and cloud, achieving higher accuracy than frontier models at up to 49x lower cost.

Reproducible Evaluation as the Substrate for Safe Prompt Compression in Production Agents

arXiv:2610.31972v1 Announce Type: new Abstract: Evaluating conversational agents that act on live, mutable enterprise data is hard: replayed sessions drift as backend state changes, and a benchmark built today can silently go stale as the data it describes moves on. We present an offline evaluation harness for a production conversational agent that stays reproducible despite the live data it operates on changing, and that can be regenerated on demand once it goes stale. We then use this harness as the fitness function for GEPA-based prompt optimization under a correctness gate that guards against reward hacking. Applied to a production merchant-facing agent, correctness-gated compression shrinks the agent's full prompt surface by 3.5-7.6% without regressing held-out task correctness, consi

In LLM evaluation, relying solely on the overall F1 score to validate automated scorers carries the risk of misleading results: a new study underscores the need to calibrate for prompt-specific differences.

An arXiv paper (2610.31678v1) points out that, in interactive evaluations of large language models, relying solely on the overall accuracy or F1 score of automated scorers can mask prompt-specific errors, leading to misjudgments of prompt effects. Empirical evidence shows that the same scorer exhibits substantial variation in its false-positive rate across different prompts (2.6%–33.3%), and a seemingly significant 9.3-point effect shrinks to 2.3 points—and becomes non-significant—under dual-LLM review.

Why it matters：This study, based on real experimental data (6 models, 180 fictional entity‑based questions, and 6 prompt conditions), reveals a widespread methodological flaw in current LLM evaluations: equating the scorer’s overall performance with its reliability on key benchmark tasks. Its central claim—that each prompt must be audited separately and uncertainty must be incorporated—rests on clear empirical evidence and offers actionable recommendations, making it highly timely (published on 2026‑10‑26) and backed by a credible source (the arXiv preprint platform).

ALTK-Evolve: Without model weight updates, relying solely on learning-based memory, it tops the τ²-bench customer service benchmark.

IBM Research has introduced the ALTK-Evolve method, which leverages GPT-4.1 to automatically extract structured behavioral guidelines—such as identity verification and refund confirmation—from its own task trajectories. These guidelines are then injected into the ReAct agent system via a purely prompt-engineering approach, significantly improving the accuracy and reliability of customer-service tasks without updating model weights or relying on human annotations. Its retail branch achieved a Pass⁴ score of 67.5%, ranking first on the τ²-bench leaderboard as of October 12, 2026, surpassing all state-of-the-art models and fine-tuned systems.

Towards Effective Process Supervision in Mathematical Reasoning

GITHUB HUGGING FACE MODELSCOPE DISCORD Introduction In recent years, Large Language Models (LLMs) have made remarkable advances in mathematical reasoning, yet they can make mistakes, such as miscalculations or logical errors, leading to wrong conclusions. Moreover, even when achieving correct final answers, these powerful models can still regularly make up plausible reasoning steps, where the final answers build upon flawed calculations or derivations, which undermine the reliability and trustworthiness of LLMs’ reasoning processes.

Introducing Qwen2-Math

GITHUB HUGGING FACE MODELSCOPE DISCORD 🚨 This model mainly supports English. We will release bilingual (English and Chinese) math models soon. Introduction Over the past year, we have dedicated significant effort to researching and enhancing the reasoning capabilities of large language models, with a particular focus on their ability to solve arithmetic and mathematical problems. Today, we are delighted to introduce a series of math-specific large language models of our Qwen2 series, Qwen2-Math and Qwen2-Math-Instruct-1.

Hello Qwen2

GITHUB HUGGING FACE MODELSCOPE DEMO DISCORD Introduction After months of efforts, we are pleased to announce the evolution from Qwen1.5 to Qwen2. This time, we bring to you: Pretrained and instruction-tuned models of 5 sizes, including Qwen2-0.5B, Qwen2-1.5B, Qwen2-7B, Qwen2-57B-A14B, and Qwen2-72B; Having been trained on data in 27 additional languages besides English and Chinese; State-of-the-art performance in a large number of benchmark evaluations; Significantly improved performance in coding and mathematics; Extended context length support up to 128K tokens with Qwen2-7B-Instruct and Qwen2-72B-Instruct.