OpenSeeker: Democratizing Frontier Search Agents by Fully Open-Sourcing Training Data

AI Can Learn Scientific Taste

MM-CondChain: A Programmatically Verified Benchmark for Visually Grounded Deep Compositional Reasoning

Can Vision-Language Models Solve the Shell Game?

OmniForcing: Unleashing Real-time Joint Audio-Visual Generation

daVinci-Env: Open SWE Environment Synthesis at Scale

Cheers: Decoupling Patch Details from Semantic Representations Enables Unified Multimodal Comprehension and Generation

LMEB: Long-horizon Memory Embedding Benchmark

DreamVideo-Omni: Omni-Motion Controlled Multi-Subject Video Customization with Latent Identity Reinforcement Learning

ShotVerse: Advancing Cinematic Camera Control for Text-Driven Multi-Shot Video Creation

Video-Based Reward Modeling for Computer-Use Agents

IndexCache: Accelerating Sparse Attention via Cross-Layer Index Reuse

Strategic Navigation or Stochastic Search? How Agents and Humans Reason Over Document Collections

Spatial-TTT: Streaming Visual-based Spatial Intelligence with Test-Time Training

Can Large Language Models Keep Up? Benchmarking Online Adaptation to Continual Knowledge Streams

ReMix: Reinforcement routing for mixtures of LoRAs in LLM finetuning

In-Context Reinforcement Learning for Tool Use in Large Language Models

MA-EgoQA: Question Answering over Egocentric Videos from Multiple Embodied Agents

Flash-KMeans: Fast and Memory-Efficient Exact K-Means

OpenClaw-RL: Train Any Agent Simply by Talking

Stepping VLMs onto the Court: Benchmarking Spatial Intelligence in Sports

InternVL-U: Democratizing Unified Multimodal Models for Understanding, Reasoning, Generation and Editing

MM-Zero: Self-Evolving Multi-Model Vision Language Models From Zero Data

Thinking to Recall: How Reasoning Unlocks Parametric Knowledge in LLMs

Omni-Diffusion: Unified Multimodal Understanding and Generation with Masked Discrete Diffusion

Geometry-Guided Reinforcement Learning for Multi-view Consistent 3D Scene Editing

CARE-Edit: Condition-Aware Routing of Experts for Contextual Image Editing

Believe Your Model: Distribution-Guided Confidence Calibration

LoGeR: Long-Context Geometric Reconstruction with Hybrid Memory

How Far Can Unsupervised RLVR Scale LLM Training?

OpenSeeker: Democratizing Frontier Search Agents by Fully Open-Sourcing Training Data

AI Can Learn Scientific Taste

MM-CondChain: A Programmatically Verified Benchmark for Visually Grounded Deep Compositional Reasoning

Can Vision-Language Models Solve the Shell Game?

OmniForcing: Unleashing Real-time Joint Audio-Visual Generation

daVinci-Env: Open SWE Environment Synthesis at Scale

Cheers: Decoupling Patch Details from Semantic Representations Enables Unified Multimodal Comprehension and Generation

LMEB: Long-horizon Memory Embedding Benchmark

DreamVideo-Omni: Omni-Motion Controlled Multi-Subject Video Customization with Latent Identity Reinforcement Learning

ShotVerse: Advancing Cinematic Camera Control for Text-Driven Multi-Shot Video Creation

Video-Based Reward Modeling for Computer-Use Agents

IndexCache: Accelerating Sparse Attention via Cross-Layer Index Reuse

Strategic Navigation or Stochastic Search? How Agents and Humans Reason Over Document Collections

Spatial-TTT: Streaming Visual-based Spatial Intelligence with Test-Time Training

Can Large Language Models Keep Up? Benchmarking Online Adaptation to Continual Knowledge Streams

ReMix: Reinforcement routing for mixtures of LoRAs in LLM finetuning

In-Context Reinforcement Learning for Tool Use in Large Language Models

MA-EgoQA: Question Answering over Egocentric Videos from Multiple Embodied Agents

Flash-KMeans: Fast and Memory-Efficient Exact K-Means

OpenClaw-RL: Train Any Agent Simply by Talking

Stepping VLMs onto the Court: Benchmarking Spatial Intelligence in Sports

InternVL-U: Democratizing Unified Multimodal Models for Understanding, Reasoning, Generation and Editing

MM-Zero: Self-Evolving Multi-Model Vision Language Models From Zero Data

Thinking to Recall: How Reasoning Unlocks Parametric Knowledge in LLMs

Omni-Diffusion: Unified Multimodal Understanding and Generation with Masked Discrete Diffusion

Geometry-Guided Reinforcement Learning for Multi-view Consistent 3D Scene Editing

CARE-Edit: Condition-Aware Routing of Experts for Contextual Image Editing

Believe Your Model: Distribution-Guided Confidence Calibration

LoGeR: Long-Context Geometric Reconstruction with Hybrid Memory

How Far Can Unsupervised RLVR Scale LLM Training?