New Model Releases
A
Showing 241–270 of 287
-
EnergyCal: Taming Loss Spikes in Muon-Family Optimizers for Sparse MoE Training
-
Exploring the Limits of the Delta Learning Hypothesis: Single-Pair Generator Biases in Direct Preference Optimization
-
V-LUMEN: Visual Lookup Memory for Embedding Scaling in Vision-Language Models
-
The State and Fate of Multilingual, Contextual Evaluation in the NLP World
-
AutoVend: Benchmarking LLM Agents on Long-Horizon Autonomous Vending Operations
-
DocXEval: A Systematic Evaluation of Explainability in Multimodal Document Understanding
-
Do We Need an LLM to Know Why Sources Conflict? Decision Models for Knowledge-Conflict Typing in RAG
-
メアドを検索→AIで“その人専用フィッシングメール”を生成する手法、クリック率が約2.7倍に 7000人以上で実験AI-personalized phishing lifts click rates about 2.7x in studyResearchers from institutions including TU Berlin published 'A Large-Scale Study of Personalized Phishing using Large Language Models', testing how effective LLM-generated spear phishing is. Searching a target's email address and generating a tailored phishing message reportedly raised click rates roughly 2.7 times in an experiment involving more than 7,000 people.
-
コスパ重視の「GPT-6.1 Terra」新登場 コスト10分の1で一部Astraと同等性能うたうOpenAI launches cost-focused GPT-6.1 Terra at one-tenth the costOpenAI has announced GPT-6.1 Terra, a new cost-efficiency-focused model. It claims performance on par with the top-tier GPT-6 Astra in some areas while costing one-tenth as much, offering a cheaper option for using high-end capabilities.
-
Expanding our partnership with GSA to bring Claude to all levels of governmentAnthropic expands GSA deal: 40% off Claude for governmentsAnthropic announced a new GSA agreement offering government partners a 40% discount on Claude Enterprise, Claude for Government and the API across all product surfaces, extended to state, local and tribal governments. A new implementation program launches too; savings are estimated at $1 billion.
-
GPT-6.1 TerraOpenAI launches GPT-6.1 Terra, fast model for visual reasoning, computer useOpenAI introduced GPT-6.1 Terra, a fast, efficient model emphasizing detailed visual reasoning, precise computer use and clearer writing, a significant step up from GPT-5.6 Terra. Simon Willison notes it sits where Sonnet does in Anthropic's lineup, and suits repetitive automated computer-use tasks.
-
IEEE Xplore Digital Library Adds AI AssistantIEEE launches AI Research Suite for Xplore digital libraryIEEE released the IEEE Xplore AI Research Suite, built on the full text of millions of peer-reviewed publications. Powered by an AI system tuned for STEM literature, it gives answers grounded in verified results, as an optional upgrade for institutional subscribers.
-
Auto-Research Arena: Benchmarking and Advancing Auto-Research in Language Modeling
-
Patient-specific rubric-based evaluation of large language models across longitudinal cancer care
-
Work Agents: Systems, Capabilities, and Optimization
-
I-HITL: Inalienable Human-in-the-Loop
-
Pretraining Knows What, Simulation Teaches How: Compositional Generalization in a Sim-Trained Robot Agent
-
FOCUS: Active Critic Learning for Long-Horizon Language Model Agents
-
Quantifying Claim Smoothing by LLM Research Agents
-
Atomic Security Concerns: Towards a Foundational Language for Security Knowledge Representation
-
A Unified Arena for LLM Agents across Diverse Eurogame
-
EEGaze: Multimodal Attention Prediction in Visual Search via EEG-Based Saliency Refinement
-
Model Selection for Off-Policy Evaluation via Pairwise Bellman Validation
-
LUCID: Local Understanding of Covariate Importance with Conditional Density Estimation
-
Preference-Guided Cooperation in Multi-Objective Multi-Agent Reinforcement Learning
-
Multi-Objective Reinforcement Learning Using Routed Ensembles and Trajectory Attribution
-
The Watchtower Imperative: Lifespan-Calibrated Evaluation for Frontier AI
-
Triage: Skip, Skim, or Refine Attention Tiles in Video Diffusion Transformers
-
Towards a Unified Framework of Optimal Input-Side Dimensional Configuration for Deep CTR Prediction
-
TrySee: Modular Perception-Enhanced VLM Grounding for Closed-Loop Robot Task Planning in Specialized Domains