LLMs
-
Google’s PaperBanana, an agentic AI that creates scientific diagrams
PaperBanana (code, paper, project page) is a multi-agent AI system that automates the…
-
TurboDiffusion makes video diffusion models 100–200× faster
TurboDiffusion (code, paper) is an acceleration framework that significantly reduces inference time and…
-
DeepSeekMath-V2, the AI that can check its own math reasoning
DeepSeekMath-V2 (paper, model) is a self-verifiable mathematical reasoning model developed by DeepSeek-AI. It…
-
MinerU2.5, a vision-language model for efficient document parsing
MinerU2.5 (paper, code) is a parsing vision-language model that converts complex documents, such…
-
AgentScope 1.0, a developer-centric framework for AI agents
AgentScope 1.0 (project page, paper, code) is a new framework designed to simplify…
-
V-JEPA 2, a model that uses videos to learn real world physics
Imagine a home robot that learns to tidy up simply by watching online…
-
FastVLM, a Vision-Language Model – CVPR 2025
Apple Research introduces FastVLM, a new architecture designed to accelerate Vision-Language Models (VLMs)…
-
AI models and book memorization, new study sparks debate
What if your favorite AI chatbot wasn’t just trained on books, but could…
-
BAGEL from ByteDance, an open-source multimodal AI
Have you ever wondered how AI can generate detailed captions from images, answer…
-
Universal geometry links multimodal embedding spaces
What if all the embedding spaces used in AI — text, images, speech,…