model releases

RMA: an Agentic System for Research-Level Mathematical Problems

arXiv:2605.22875v1 Announce Type: new Abstract: We present extbf{Research Math Agents (RMA)}, an agentic framework for automated reasoning on research-level mathematical problems. Unlike prior studies

Model Releases

RoboSurg-VQA: A Multimodal Benchmark for Surgical Segmentation-Aware Visual Question Answering

DGX agent

arXiv:2605.23068v1 Announce Type: new Abstract: Reliable visual understanding in robot-assisted and minimally invasive surgery (RMIS/MIS) demands more than accurate masks: in clinical practice, clinic

model-releasesarxiv-cs-cv

25 May 2026

Same Model, Different Weakness: How Language and Modality Reshape the Jailbreak Attack Surface in Frontier MLLMs

arXiv:2605.23157v1 Announce Type: new Abstract: The attack surface of a multimodal large language model (MLLM) is language-dependent in ways that reveal the mechanistic structure of alignment failures

SciAtlas: A Large-Scale Knowledge Graph for Automated Scientific Research

arXiv:2605.22878v1 Announce Type: new Abstract: The exponential growth of global academic output has confronted researchers and AI agents with an unprecedented ``information explosion,'' where fragmen

SciHorizon-GENE: Benchmarking LLM for Life Sciences Inference from Gene Knowledge to Functional Understanding

arXiv:2601.12805v3 Announce Type: replace-cross Abstract: Large language models (LLMs) have shown growing promise in biomedical research, particularly for knowledge-driven interpretation tasks. Howeve

Seeing without Looking: Do Vision-Language Benchmarks Really Test Vision?

arXiv:2605.22903v1 Announce Type: cross Abstract: Benchmark accuracy is often implicitly assumed to reflect grounded visual understanding in vision-language models (VLMs), yet it remains unclear to wh

Semantically Structured Mixture-of-Experts for Compositional Robotic Manipulation

model-releasesarxiv-cs-ro

arXiv:2605.23477v1 Announce Type: new Abstract: Diffusion-based policies have established a new standard for precise robotic manipulation but face a critical scalability bottleneck: high-performance m

SemEval-2026 Task 6: CLARITY -- Unmasking Political Question Evasions

arXiv:2603.14027v2 Announce Type: replace Abstract: Political speakers often avoid answering questions directly while maintaining the appearance of responsiveness. Despite its importance for public di

SkillOpt: Executive Strategy for Self-Evolving Agent Skills

arXiv:2605.23904v1 Announce Type: new Abstract: Agent skills today are hand-crafted, generated one-shot, or evolved through loosely controlled self-revision, none of which behaves like a deep-learning

Sparse Autoencoders Map Brain-LLM Alignment onto Cortical Semantic Topography

arXiv:2605.23035v1 Announce Type: cross Abstract: Intermediate layers of large language models (LLMs) best predict human brain responses to language, one of the most robust findings in computational n

Speak-to-Structure: Evaluating LLMs in Open-domain Natural Language-Driven Molecule Generation

arXiv:2412.14642v4 Announce Type: replace Abstract: Recently, Large Language Models (LLMs) have demonstrated great potential in natural language-driven molecule discovery. However, existing datasets a

STAMBRIDGE: Spectral-Temporal Amplitude-aware Mid-Feature Bridge for EEG Visual Decoding

arXiv:2605.23137v1 Announce Type: cross Abstract: Electroencephalography (EEG) visual decoding remains challenging due to the modality gap between low-SNR neural signals and highly structured vision--

StereoGenBench: A Synthetic Multi-Camera Benchmark for Stereo Generation under Controlled Baseline Regimes

arXiv:2605.23237v1 Announce Type: new Abstract: Stereo image and video generation, stereo geometry estimation, and condition-controlled view synthesis require paired data in which the variables that d

Strategic Coercion Within Alliances: The Greenland Sovereignty Game as an AI Stress Test

arXiv:2605.22841v1 Announce Type: cross Abstract: What happens when the strongest alliance member pressures a weaker member over territory and strategic control? We examine the Greenland sovereignty c

Tabular PDF Information Extraction with Local LLMs and Layout-Aware Parsing: A Reliability Evaluation

arXiv:2604.00003v2 Announce Type: replace-cross Abstract: Extracting structured information from academic PDF documents is non trivial: a single page typically combines free text metadata with tabular

Targeted Regularization for Causal Effect Estimation with Exponential Dispersion Family Outcomes

model-releasesarxiv-cs-lg

arXiv:2502.07295v2 Announce Type: replace Abstract: Neural Networks (NNs) for causal effect estimation have shown strong empirical performance, yet endowing them with desirable semiparametric properti

TEAM: Temporal-Spatial Consistency Guided Expert Activation for MoE Diffusion Language Model Acceleration

arXiv:2602.08404v2 Announce Type: replace Abstract: Diffusion large language models (dLLMs) have recently gained significant attention due to their inherent support for parallel decoding. Building on

The Misattribution Gap: When Memory Poisoning Looks Like Model Failure in Agentic AI Systems

arXiv:2605.22842v1 Announce Type: cross Abstract: Multi-agent AI pipelines typically assume that agent misconduct originates from model misalignment. We identify a structural failure in this assumptio

The Readout Shortcut: Positional Number Copying Dominates Arithmetic CoT Readout in Small Language Models

arXiv:2605.22870v1 Announce Type: cross Abstract: Chain-of-thought (CoT) prompting is necessary for arithmetic in small language models, yet shuffling its steps preserves most performance. What does C

The Surprising Difficulty of Search in Model-Based Reinforcement Learning

arXiv:2601.21306v2 Announce Type: replace-cross Abstract: This paper investigates search in model-based reinforcement learning (RL). Conventional wisdom holds that long-term predictions and compoundin

Transcoders Trace Visual Grounding and Hallucinations in Vision-Language Models

arXiv:2605.22902v1 Announce Type: cross Abstract: Generative Vision-Language Models (VLMs) perform well on multimodal reasoning, but how visual inputs are transformed to text remains poorly understood

Understanding and Improving Noisy Embedding Techniques in Instruction Finetuning

arXiv:2605.23171v1 Announce Type: cross Abstract: Recent advancements in instructional fine-tuning have injected noise into embeddings, with NEFTune (Jain et al., 2024) setting benchmarks using unifor

Unextractable Protocol Models: Collaborative Training and Inference without Weight Materialization

model-releasesarxiv-cs-lg

arXiv:2605.23464v1 Announce Type: new Abstract: We consider a decentralized setup in which the participants collaboratively train and serve a large neural network, and where each participant only proc

Using Ensemble Diffusion to Estimate Uncertainty for End-to-End Autonomous Driving

arXiv:2506.00560v2 Announce Type: replace-cross Abstract: End-to-end planning systems for autonomous driving are rapidly improving, especially in closed-loop simulation environments like CARLA. Many s

VDE: Training-Free Accelerating Rectified Flow Model via Velocity Decomposition and Estimation

arXiv:2605.23381v1 Announce Type: new Abstract: Though rectified flow models have achieved remarkable performance in image, video, and 3D generation, their practical deployments are challenged by slow

Vector Retrieval with Similarity and Diversity: How Hard Is It?

arXiv:2407.04573v4 Announce Type: replace-cross Abstract: Dense vector retrieval is an important building block of modern machine learning systems, underlying applications ranging from semantic search

VideoOdyssey: A Benchmark for Ultra-Long-Context and Omni-Modal Video Understanding

arXiv:2605.22907v1 Announce Type: new Abstract: Real-world long video understanding requires models to perform continuous tracking, information integration and memory retention over massive temporal s

VideoTemp-o3: Harmonizing Temporal Grounding and Video Understanding in Agentic Thinking-with-Videos

arXiv:2602.07801v4 Announce Type: replace-cross Abstract: In long-video understanding, conventional uniform frame sampling often fails to capture key visual evidence, leading to degraded performance a

VINS-120K: Ultra High-Resolution Image Editing with A Large-Scale Dataset

arXiv:2605.23518v1 Announce Type: new Abstract: Directly editing ultra-high-resolution (UHR) images is valuable but underexplored, primarily due to the lack of high-quality data and the challenge in m

VisAnalog: A Diagnostic Suite for Visual Concept Transfer on Natural Images

arXiv:2605.23141v1 Announce Type: new Abstract: A useful test of visual concept learning is not just whether a model can recognize a concept in a single image, but whether it can preserve and manipula

What Linear Probes Miss: Multi-View Probing for Weight-Space Learning

arXiv:2605.23410v1 Announce Type: cross Abstract: The explosive growth of open-source model repositories has created a Model Jungle, where checkpoints are frequently shared without adequate documentat

What Training Data Teaches RL Memory Agents: An Empirical Study of Curriculum Effects in Memory-Augmented QA

arXiv:2605.23067v1 Announce Type: new Abstract: Reinforcement learning (RL) has emerged as a viable recipe for training LLM agents to reason over external memory banks in multi-session dialogue. Exist

When Good Equations Get Bad Scores: Improving Symbolic Regression Through Better Parameter Optimization

arXiv:2605.23272v1 Announce Type: cross Abstract: Symbolic Regression (SR) plays a central role in scientific knowledge discovery by distilling mathematical equations from observational data. Most exi

When Symptoms Are Not Enough: Evidence-Weighting Patterns in Large Language Model Psychiatric Screening

arXiv:2605.23148v1 Announce Type: new Abstract: As demand for mental health care outpaces clinician-delivered assessment, scalable screening tools are increasingly needed. Large language models (LLMs)

Wordle 1,801 4/6 🟨🟨⬛⬛⬛ ⬛🟩⬛⬛⬛ ⬛🟩🟩🟨⬛ 🟩🟩🟩🟩🟩

model-releasesanthropic--x

This post documents a Wordle game result where the player solved puzzle #1,801 in four attempts using color-coded feedback (yellow for correct letters in wrong positions, green for correct letters in

XAttnMark: Learning Robust Audio Watermarking with Cross-Attention

arXiv:2502.04230v3 Announce Type: replace-cross Abstract: The rapid proliferation of generative audio synthesis and editing technologies has raised serious concerns about copyright infringement, data

🇺🇸🇪🇺 A new study from the US-based New England Journal of Medicine found that Americans die earlier across all income levels compared to…

model-releasesyann-lecun--x

🇺🇸🇪🇺 A new study from the US-based New England Journal of Medicine found that Americans die earlier across all income levels compared to their European counterparts. What’s especially notable is that

Aleph 2.0 will blow your mind

model-releasescristobal-valenzuela--x

Aleph 2.0 will blow your mind Just tested Runway Aleph 2.0 and this blew my mind a bit lol I saw a video like this when Aleph first released and with the new Aleph 2.0 I wanted to create my own versio

An interesting work on Physical AI: PhysX-Omni. First unified sim-ready generation framework for rigid, deformable, and articulated objects,…

model-releasesclem-delangue--x

An interesting work on Physical AI: PhysX-Omni. First unified sim-ready generation framework for rigid, deformable, and articulated objects, with a diverse dataset and new benchmark. 🌐 https://physx-o

Built an AI screen memory using llama.cpp + Gemma 4 — remembers everything you do on your computer,search/chat or make agents over it. 100% local

This project demonstrates a local AI system built with llama.cpp and Gemma 4 that captures and analyzes screen activity to create persistent memory of user computer interactions, enabling search, chat

model-releasesr-ollama

DeepSeek says it will lower V4 Pro API prices by 75% to 0.435/1M input and 0.87/1M output tokens, making permanent the discount prices set to expire on May 31 (Bloomberg)

Bloomberg: DeepSeek says it will lower V4 Pro API prices by 75% to 0.435/1M input and 0.87/1M output tokens, making permanent the discount prices set to expire on May 31 — DeepSeek said it will make p

model-releasestechmeme

It makes many online spaces intolerable. If I want to talk to ChatGPT or Claude, I'll just talk to ChatGPT or Claude, I don't need to talk t…

model-releasesethan-mollick--x

It makes many online spaces intolerable. If I want to talk to ChatGPT or Claude, I'll just talk to ChatGPT or Claude, I don't need to talk to ChatGPT and Claude pretending to be DoofWarrior123 on X wi

It’s no longer just AI companies & their founders being sued over AI training - individual researchers are now being sued, too. In a new law…

model-releasesgary-marcus--x

It’s no longer just AI companies & their founders being sued over AI training - individual researchers are now being sued, too. In a new lawsuit, two authors allege that Guillaume Lample, while an AI

Mad House — Usborne Creepy Computer Games

model-releasessimon-willison

Tool: Mad House — Usborne Creepy Computer Games Via Hacker News I learned that UK publisher Usborne published free PDFs of their 1980s Computer Books, some of which I remember working through on my Co

People often ask what my biggest tip is for getting the most out of Claude Code. These days my #1 tip is: use auto mode Auto mode means no m…

model-releasesboris-cherny--x

People often ask what my biggest tip is for getting the most out of Claude Code. These days my #1 tip is: use auto mode Auto mode means no more permission prompts. It is the key building block for mul

quick summary of someone's github, cool! here's me

model-releasesyohei-nakajima--x

quick summary of someone's github, cool! here's me I always wanted a GitHub dashboard: See my repos, open Issues/PRs, what version I released last, how many commits since last release. So I built one

Wordle 1,799 4/6 ⬛⬛⬛⬛⬛ ⬛⬛⬛⬛⬛ ⬛🟨⬛⬛⬛ 🟩🟩🟩🟩🟩

model-releasesanthropic--x

This post shows a Wordle game result where the player solved puzzle #1,799 in 4 attempts, with the final answer being a five-letter word where all letters are in the correct positions (indicated by th

Wordle 1,800 4/6 ⬛⬛⬛⬛🟩 ⬛🟩🟨⬛⬛ 🟩🟩🟨⬛🟩 🟩🟩🟩🟩🟩

model-releasesanthropic--x

This post documents a Wordle game result where the player solved puzzle #1,800 in 4 attempts, using the color-coded feedback system (gray for wrong letters, yellow for correct letters in wrong positio