New Model Releases
A
Showing 31–60 of 289
-
How Much Can Tool Calls Reveal About You? Privacy Leakage from Tool-Call Traces
-
AI evaluators favor job applications written by certain AI models, independent of candidate quality
-
Learning to Interact and Direct: Co-Evolving Video Director Agents
-
SimXAI: A Simple Drop-in Layer Increases Interaction Quality and Understanding in Conversational XAI
-
Overcoming the Representational Limitations of Amplitude Encoding for Quantum Neural Networks
-
ElTab: Elastically Looped Tabular Transformers
-
DIFFNAS: Predictor-Free Neural Architecture Search via Diffusion Model and Zero-Cost Evaluation
-
Accurate and Auditable ECG Classification via Visibility Graph Neural Networks
-
Peeling the Matryoshka: Extracting Ontological Structure from Retrieval Embeddings
-
Stepwise-Querying: Distilling Reasoning without Access to Teacher Chain-of-Thought
-
Multimodal In-Context Learning with Explicit Rule Induction
-
Simple Features Beat Most AI Control Monitors
-
Advancing Clinical Trial Outcome Prediction with Domain-Specific Large Language Models
-
GraphNIAH: A Benchmark for Long-Context Reasoning over Real-World Graphs
-
Does Reasoning Make Language Models Agree?
-
キオクシア中間決算「過去最高」連呼 純利益2.2兆円、初の配当実施 AI需要で「成長さらに加速」Kioxia posts record H1 net profit of 2.2T yen, first dividendKioxia Holdings reported interim results for the fiscal year ending March 2027, with net profit of 2.2 trillion yen. Its three-month second quarter alone exceeded the prior full-year revenue, and AI demand is expected to accelerate growth further. It will pay its first dividend since listing.
-
GRIT: Gradient Masking for Visually Misgrounded Tokens in Text-Rich Reasoning
-
BYOD: Build Your Own DLM - Efficient Autoregressive-to-Diffusion Language Model Conversion
-
Enhancing Knowledge Entity Relation Extraction in Abstracts and Full-text Content of Academic Articles Using Large Language Models
-
DAPS: Dependency-Aware Premise Selection for LLM Theorem Proving
-
“でぶねこ”の重み公開 1兆パラメータのAIモデル「Mistral Large 4」（Le Chonk）がダウンロード可能にMistral releases weights for 1T-parameter Mistral Large 4France's Mistral AI released the weights of Mistral Large 4, nicknamed Le Chonk, on Hugging Face for download. The 1-trillion-parameter model reportedly performs strongly in coding and cybersecurity.
-
BlindToChange: AI Agents Can Find Changes When Asked But Do They Look When Needed?
-
Hybrid Computer Use: A Unified Programmatic, Semantic, and Visual Interface for General Agents
-
SparseGDN: Chunk Sparse in Linear Attention
-
Gemini Notebook、無料で数百冊の教科書にアクセス可能に 米国・インドの出版社提供 日本でも利用可Gemini Notebook offers hundreds of textbooks free, including in JapanGoogle announced that hundreds of textbooks supplied by providers in the United States and India can now be accessed for free within Gemini Notebook. The textbook access is also available to users in Japan.
-
ProSearch: Benchmarking Multi-Constraint Protocol Retrieval in Experimental Science
-
What We Keep of What We Forget: Replacement Operators and Routing Semantics in Training-Free MoE Expert Pruning
-
SERIAL-7: Deriving, Realizing, and Preserving Machine-Verifiable Outputs Across Representations
-
From Transcription to Reformulation: Defining and Benchmarking Grounded Generative Speech Recognition
-
Meta Ray-Banで向かいの席から撮影→指の動きからスマホのパスワードを解読する盗み見攻撃Smart-glasses video of finger movements can crack phone passwordsTel Aviv University researchers presented 'Watch and Crack,' a paper showing that passwords can be inferred from finger movements recorded by smart glasses. The shoulder-surfing attack films from across the table with Meta Ray-Ban glasses to read smartphone passwords.