CompanyAnthropic8 recent entries16 Oct 2026Screwtape: Adversarial Transcript Editing for Monitor Red-TeamingarXiv:2610.20941v1 Announce Type: cross Abstract: We present Screwtape, a pipeline that converts real agent transcripts into attacks that evade action-only monitors. An LLM attacker edits a transcript→16 Oct 2026Librarian Agents: Training LLMs to Curate Long-Context Workspaces that Anticipate Future QueriesarXiv:2610.22273v1 Announce Type: cross Abstract: Reasoning over long document contexts remains a challenge for large language models (LLMs). Standard retrieval methods are inherently reactive, bottle
CompanyOpenAI8 recent entries14 Oct 2026I built a cheap dual 3090 rig from used parts, but why isn't it much faster?Hey everyone, I recently put together a local LLM machine using mostly second hand hardware. I wanted to experiment with local models for app development without spending a fortune on a new workstatio→15 Oct 2026UDHRBench: Benchmarking Human Rights in Large Language ModelsarXiv:2610.15179v1 Announce Type: cross Abstract: Large language models are increasingly deployed in contexts requiring adherence to ethical and legal norms, yet their alignment with internationally r→15 Oct 2026LLMRec: Benchmarking Large Language Models on Recommendation TaskarXiv:2308.12241v2 Announce Type: replace-cross Abstract: Recently, the fast development of Large Language Models (LLMs) such as ChatGPT has significantly advanced NLP tasks by enhancing the capabilit→15 Oct 2026EmoXpt: Analyzing Emotional Variances in Human Comments and LLM-Generated ResponsesarXiv:2501.06597v1 Announce Type: cross Abstract: The widespread adoption of generative AI has generated diverse opinions, with individuals expressing both support and criticism of its applications. T→15 Oct 2026CWE-bench: Refocusing Agentic Security from Exploitation to DefensearXiv:2610.17020v1 Announce Type: cross Abstract: Cybersecurity evaluations of AI agents have focused largely on offense, yet a working exploit does not show that an agent can make software safe. An a→15 Oct 2026A large-scale pipeline for LLM-assisted corpus annotation: variation and change in the English consider constructionarXiv:2510.12306v4 Announce Type: replace Abstract: As natural language corpora expand at an unprecedented rate, manual annotation remains a significant methodological bottleneck in corpus linguistic →16 Oct 2026System1-Arena: Which Decisions Can Be Left to System 1?arXiv:2610.21244v1 Announce Type: cross Abstract: Large language models (LLMs) often make discrete decisions, such as flagging phishing or escalating to a human. System-1 models, such as Jev and the O→16 Oct 2026AI Consilium: a Benchmark of LLM Clinical Multidisciplinary Second OpinionsarXiv:2610.22298v1 Announce Type: cross Abstract: Large language models (LLMs) can generate differential diagnoses for complex cases, but a useful second opinion must also be factually grounded, clini
CompanyGoogle8 recent entries16 Oct 2026INFUSE: A Unified Surface Benchmark for Indirect Prompt Injection in Agentic SystemsarXiv:2610.20776v1 Announce Type: cross Abstract: LLM-based agents increasingly execute long-horizon, end-to-end workflows involving shell commands, code modification, dependency installation, and int→16 Oct 2026Four evaluation engineering lessons learned from building agent pluginsThe real challenge of building an agentic system isn't the construction, it's the measurement. Evaluation engineering is a critical foundation for building reliable AI systems. Setting up systematic, →16 Oct 2026Day 0 support for Reflection’s Beam on Google Cloud: 5x faster perf with llm-d and NVIDIA DynamoToday, Reflection released Beam, its first open-weight model built to give developers, researchers, and enterprises open control and customization for frontier software engineering and long-horizon ag→16 Oct 2026ChaosBench-Logic: A Benchmark for Logical and Symbolic Reasoning on Chaotic Dynamical SystemsarXiv:2601.01982v2 Announce Type: replace Abstract: Large language models (LLMs) excel at natural language tasks but remain brittle in domains requiring precise logical and symbolic reasoning. Chaotic→16 Oct 2026Beyond Generation: Using Multimodal LLMs as Detectors of LLM-Generated Deepfake Facial ImagesarXiv:2610.21760v1 Announce Type: new Abstract: As image generation models become increasingly capable of producing highly realistic content, traditional forensic techniques are struggling to keep up.→16 Oct 2026AutoOnboarder: Autonomously Onboarding LLM AgentsarXiv:2610.21218v1 Announce Type: new Abstract: Agentic large language models have been documented to lose reasoning capability when placed inside a harness, the software that mediates their observati→16 Oct 2026Architecting autonomy: A design pattern for high-compute, governed agents in biotechIn production systems today, most AI agents operate as lightweight dispatchers. They parse a prompt, make an API call, retrieve a record, and hand back control. Their execution is fast, stateless and →16 Oct 2026A team of Gemini 4 Argon agents beats MinecraftarXiv:2610.22319v1 Announce Type: new Abstract: We placed four Gemini 4 Argon agents in a sandboxed Minecraft survival world and gave them one goal: defeat the Ender Dragon. Each agent controlled a pl
CompanyMeta8 recent entries16 Oct 2026INFUSE: A Unified Surface Benchmark for Indirect Prompt Injection in Agentic SystemsarXiv:2610.20776v1 Announce Type: cross Abstract: LLM-based agents increasingly execute long-horizon, end-to-end workflows involving shell commands, code modification, dependency installation, and int→16 Oct 2026Fractional Power-of-Two Quantization for Efficient and Effective Multiplier-Free LLM InferencearXiv:2610.21257v1 Announce Type: cross Abstract: Large Language Models (LLMs) demand efficient inference, where the multiply-accumulate (MAC) operations in linear projections dominate compute and ene→16 Oct 2026FIRM: Fisher-Informed Retention Merging for Federated LLM UnlearningarXiv:2610.21720v1 Announce Type: cross Abstract: Federated LLM unlearning addresses targeted knowledge removal when forget and retain data remain distributed across private clients and are inaccessib→16 Oct 2026FedSkillGit: Federated Skill Evolution via Version ControlarXiv:2610.21348v1 Announce Type: new Abstract: Large language model agents can improve their capabilities by evolving reusable natural-language skills from execution experience. In real-world deploym→16 Oct 2026ChaosBench-Logic: A Benchmark for Logical and Symbolic Reasoning on Chaotic Dynamical SystemsarXiv:2601.01982v2 Announce Type: replace Abstract: Large language models (LLMs) excel at natural language tasks but remain brittle in domains requiring precise logical and symbolic reasoning. Chaotic→16 Oct 2026CAT-alyzing Honesty: Continuous Adversarial Training for Transferable LLM HonestyarXiv:2610.22292v1 Announce Type: cross Abstract: Large language models (LLMs) are becoming prevalent in scenarios where honesty is critical. However, despite alignment training, these models still ex→16 Oct 2026CARE-SD: Explainable Reasoning Architecture for Local SLM Self-Disclosure Analysis in Textual DataarXiv:2610.21151v1 Announce Type: cross Abstract: Cybersecurity interviews uncover critical security experiences, behaviors, and privacy concerns often not fully captured by conventional technical ass→16 Oct 2026Agent Skill Framework: Perspectives on the Potential of Small to Medium Language Models in Industrial EnvironmentsarXiv:2602.16653v4 Announce Type: replace Abstract: Skills are getting increasingly used in recent agentic frameworks, performing robustly with proprietary LLMs. Yet, their effectiveness for small and
CompanyMistral8 recent entries14 Oct 2026LIFT: Likelihood-Based Intervention Faithfulness TrainingarXiv:2509.13334v2 Announce Type: replace-cross Abstract: Chain-of-thought (CoT) reasoning improves LLM accuracy on complex tasks, but CoTs are often unfaithful; that is, the model does not rely on th→14 Oct 2026GA-SVD: Generation-aware Singular Value Decomposition for LLM CompressionarXiv:2610.15398v1 Announce Type: new Abstract: Large language models (LLMs) demonstrate strong generalization across a broad range of reasoning and generation tasks, but their large parameter counts →14 Oct 2026Beyond Adjacent Layers: Graph-Guided Layer Fusion for Compressing Large Language ModelsarXiv:2610.15159v1 Announce Type: new Abstract: Existing depth-compression methods for large language models (LLMs) score layers individually or group only adjacent or contiguous layers, implicitly as→15 Oct 2026SpectrumQuant: Spectrum Alignment for LLM QuantizationarXiv:2610.20303v1 Announce Type: new Abstract: Orthogonal transforms have emerged as an effective tool for low-bit quantization of large language models (LLMs), as they reshape activation distributio→15 Oct 2026AutoSec-RAN: Specification-Grounded LLM Fine-Tuning for Automated Security Analysis of O-RANarXiv:2610.15995v1 Announce Type: cross Abstract: This paper presents AutoSec-RAN, a specification-grounded large language model (LLM) framework for automated security triage of Open Radio Access Netw→15 Oct 2026AlignMerge - Alignment-Preserving Large Language Model Merging via Fisher-Guided Geometric ConstraintsarXiv:2512.16245v2 Announce Type: replace Abstract: Merging large language models (LLMs) is a practical way to compose capabilities from multiple fine-tuned checkpoints without retraining. Yet standar→16 Oct 2026Towards Emotion Consistency Analysis of Large Language Models in Emotional Conversational ContextsarXiv:2605.06476v2 Announce Type: replace Abstract: In this work, we conduct an analysis to examine the consistency of Large Language Models (LLMs) with respect to their own generated responses in an →16 Oct 2026Human Interventions as Labels: Localizing, Steering, and (Sometimes) Editing the Computations Behind LLM Agent FailuresarXiv:2610.21708v1 Announce Type: cross Abstract: When an LLM agent misbehaves, a human steps in, and that intervention is usually discarded or replayed as training data. We treat it instead as a meas
CompanyxAI8 recent entries9 Oct 2026WorldBench: Evaluating LLMs on Three.js Voxel World GenerationarXiv:2610.10622v1 Announce Type: cross Abstract: Large language models can now write complete, interactive 3D worlds as code, but grading those worlds automatically is unreliable. Existing judges tak→12 Oct 2026People Making Conversation: Evaluating How Naturally Speech Models ConversearXiv:2610.12802v1 Announce Type: new Abstract: A conversational speech model has to decide, at every pause, whether to speak. Existing evaluations test boundary detection, policy adherence, or aggreg→12 Oct 2026Over the last few months at @river_ai_inc we’ve developed what we believe is the best system in the world for building your own internal cus…Over the last few months at @river_ai_inc we’ve developed what we believe is the best system in the world for building your own internal custom models, driven by 10 years of experience training top mo→12 Oct 2026🇫🇷 Le Chonk joins the sudo L7 leaderboard 🇫🇷 We just added @MistralAI's Mistral Large 4 to sudo L7, our benchmark for staff-level softwa…🇫🇷 Le Chonk joins the sudo L7 leaderboard 🇫🇷 We just added @MistralAI's Mistral Large 4 to sudo L7, our benchmark for staff-level software engineering. Mistral Large 4 scores 13.3%, ahead of Muse Spar→15 Oct 2026UDHRBench: Benchmarking Human Rights in Large Language ModelsarXiv:2610.15179v1 Announce Type: cross Abstract: Large language models are increasingly deployed in contexts requiring adherence to ethical and legal norms, yet their alignment with internationally r→15 Oct 2026REPRO: Reflective Evolutionary Prompt Optimizer for Automated LLM Red-TeamingarXiv:2610.15513v1 Announce Type: cross Abstract: Safety-aligned large language models (LLMs) remain vulnerable to adversarial prompts that bypass their safety training. Existing automated red-teaming→15 Oct 2026MolHop: Scaffold-Hopping Molecular Design with Agentic Reinforcement LearningarXiv:2610.19920v1 Announce Type: cross Abstract: Molecular design is often formulated as an optimization problem over existing molecular spaces. However, molecular optimization and the discovery of s→15 Oct 2026Can agents whose sole duty is to keep humans alive be hijacked to kill? Delayed multi-turn derived-action trap attacks in simulated high-stakes environmentsarXiv:2610.15153v1 Announce Type: new Abstract: We ask whether a tool-using agent whose sole duty is to keep humans alive can be induced, by delayed prompt injection, to kill. The keep-alive duty is i
CompanyDeepSeek8 recent entries15 Oct 2026Beyond Outcome Feedback: Trajectory-Quality-Guided Experience-based LearningarXiv:2610.14877v1 Announce Type: new Abstract: Experience-based learning enables large language model (LLM) agents to improve over time by reusing knowledge from past executions. However, existing me→15 Oct 2026AtomSearch: Evidence-Grounded Tree Search for Agent Skill OptimizationarXiv:2610.19951v1 Announce Type: new Abstract: Agent skills adapt frozen models through editable external instructions. Revising several rules at once makes changes difficult to inspect, while judgin→16 Oct 2026What Coding Agents Leave Unsaid: Measuring Task-State Gaps Between Execution and ReportsarXiv:2610.20003v2 Announce Type: replace Abstract: Human supervisors rely on coding agents' reports to assess progress, verify repairs, and decide when to intervene. But a supervisor cannot investiga→16 Oct 2026MemScope: Evolving Agent Memory Architectures through Localized Code RepairarXiv:2610.21170v1 Announce Type: new Abstract: Agent memory architectures define how LLM agents store and reuse experience across tasks, but they are typically fixed after deployment, limiting their →16 Oct 2026Large Language Models Naively Recover Ethnicity from Individual RecordsarXiv:2601.21132v2 Announce Type: replace Abstract: I demonstrate that large language models can infer ethnicity from names with accuracy exceeding that of Bayesian Improved Surname Geocoding (BISG) w→16 Oct 2026ISW-Agent: Integrating LLM-Driven Dialogue with n8n Workflow for Internal Solitary Wave Remote Sensing Image ProcessingarXiv:2610.21502v1 Announce Type: new Abstract: Internal solitary wave (ISW) remote sensing requires task parameters and spatial information to be transferred across image acquisition, detection, stri→16 Oct 2026IneqAgent: Agentic Inequality Proving in Lean with Symbolic ComputationarXiv:2610.20681v1 Announce Type: new Abstract: Formal inequality proving requires mathematical reasoning and exact computation, often connected through intermediate lemmas and goal reformulations. We→16 Oct 2026HTN-Orchestrate: A Hierarchical Task Network Framework for Multi-Agent OrchestrationarXiv:2610.22195v1 Announce Type: new Abstract: Orchestrating multiple agents on a complex task requires reconciling local subtask execution with global task planning. Dynamic dispatch approaches prov
CompanyNVIDIA8 recent entries15 Oct 2026A Sovereign, Indic-First, Reasoning ModelarXiv:2610.15371v1 Announce Type: cross Abstract: Gnani-Evon-v3.3-30B-A3B is a long-context hybrid Mamba-Transformer mixture-of-experts model for reasoning, instruction following, and tool use in Engl→15 Oct 20265/5 Across the board, results show the business impact and differentiation of sovereign specialized intelligence, and the opportunity for co…5/5 Across the board, results show the business impact and differentiation of sovereign specialized intelligence, and the opportunity for companies to reach AI independence. Thanks to @nvidia, @Gong_i→16 Oct 2026RTX 2080 Ti 22GB modded card for local LLM inference – anyone tested one?Title: I'm building a small local AI server to validate a project I've developed. The application currently runs Qwen 9B and the server will only be used for an initial pilot with a few users. I have →16 Oct 2026Priming Base Models for Agentic Post-Training: An Open Midtraining RecipearXiv:2610.22019v1 Announce Type: cross Abstract: Frontier models, general-purpose and domain-specific alike, now include a midtraining stage between pretraining and post-training, but little has been→16 Oct 2026Day 0 support for Reflection’s Beam on Google Cloud: 5x faster perf with llm-d and NVIDIA DynamoToday, Reflection released Beam, its first open-weight model built to give developers, researchers, and enterprises open control and customization for frontier software engineering and long-horizon ag→16 Oct 2026BitSparse: Compact Activation Storage for ReLU Feed-Forward NetworksarXiv:2610.22256v1 Announce Type: new Abstract: Reducing memory use in large language model (LLM) training is an ongoing challenge. Activation checkpointing saves memory by recomputation, while lossy →16 Oct 2026Architecting autonomy: A design pattern for high-compute, governed agents in biotechIn production systems today, most AI agents operate as lightweight dispatchers. They parse a prompt, make an API call, retrieve a record, and hand back control. Their execution is fast, stateless and →16 Oct 2026America’s next top model, Reflection’s Beam, is now live on Together AI. Trained on Together GPU Clusters, Beam is Reflection’s first open-w…America’s next top model, Reflection’s Beam, is now live on Together AI. Trained on Together GPU Clusters, Beam is Reflection’s first open-weight model. • ~501B-parameter MoE, 23B active • Built for c