Wiredframe Radar logo

Category intelligence

Research Briefing — July 30, 2026

129 current items analyzed and ranked.

Executive synthesis

Research Summary

Today's research highlights a strategic pivot toward native foundational memory architectures, high-frequency edge execution for robotics, real-time spatial world models, and critical frontier safety milestones.

Autonomous R&D & Memory Foundations

  • Open-Ended AI Research Evaluation: Establishes a concrete methodology for measuring progress toward fully automated AI scientific discovery through double-blind shadow evaluations graded by original study authors.
  • Metis: Replaces external vector stores with persistent backbone memory states embedded natively within foundation model architectures, eliminating retrieval latency and improving long-horizon reasoning context.

Embodied AI & High-Frequency Robotics

Interactive World Models & Video Generation

  • Wonder: Leverages dense coordinate fields and dynamic memory retrieval for real-time, camera-controllable video world models geared toward long-horizon spatial navigation.
  • Visko Orbis 1.0: Achieves continuous 4K resolution interactive video generation over hour-scale rollouts through dynamic prompt injection and multi-scale bounded memory.

Unified Multimodality & Safety Governance

Key Themes

Robotics & Embodied AI · 14World Models & Video Generation · 12AI Agents & Memory Systems · 15Multimodal AI & Perception · 13Reinforcement Learning & Distillation · 10AI Safety & Alignment · 10Computer Vision & 3D Reconstruction · 11Mechanistic Interpretability & Model Analysis · 4Healthcare & Medical AI · 8Generative Models & Multimodal Systems · 7

Primary evidence

Top Ranked Signals

Research AlphaXiv Trending Jul 29

Can AI agents conduct open-ended AI research? Early evidence from two case studies

By Peter Kirgis, Sayash Kapoor, Andrew Schwartz, Stephan Rabanser, David Africa, Konstantinos Voudouris, Viet Nguyen, Toby Pilditch, Magda Dubois, Harry Coppock, Cozmin Ududec, Nitya Nadgir, Matilda Orona, Tilman Bayer, Derrick Chan-Sew, Yue Ling, Abhishek Shetty, Helen Toner, Gillian Hadfield, Seth Lazar, Steve Newman, Shoshannah Tekofsky, Rishi Bommasani, Arvind Narayanan

90 score
AI Analysis

Uses shadow evaluations where original authors grade frontier AI agents attempting open-ended research questions from unpublished papers.

Forecasts of explosive AI progress hinge on AI agents automating AI research. But evidence on whether agents can carry out open-ended AI research is thin. Current evaluations either test agents on narrow, verifiable tasks, which excludes open-ended research, or submit AI-generated papers to blind peer review, which is overstretched, stochastic, and suffers from poor review quality. We introduce a third way to measure progress towards AI R\&D automation. An agent takes on the central, open-ended
AI ResearchEvaluation
Research AlphaXiv Trending Jul 29

Metis: Memory Foundation Model

By Zeyu Zhang, Ziliang Guo, Yihang Sun, Xichong Zhang, Xixuan Hao, Zehao Lin, Yang Zhang, Xiaoyan Zhao, Tong Shen, Bo Tang, Zhi-Qin John Xu, Junchi Yan, Haofen Wang, Xu Chen, Feiyu Xiong, Zhiyu Li, Tat-Seng Chua

89 score
AI Analysis

Introduces memory foundation models (Metis) featuring persistent backbone memory states and native memory procedures.

Recent advances in AI agents have increasingly internalized native capabilities into their underlying foundation models, giving rise to multimodal foundation models and large reasoning models. However, agent memory is still primarily implemented through external modules, leaving the native memory capability largely unexplored. In this paper, we take a first step toward this direction by introducing memory foundation models, which empower foundation models with native memory capabilities. We form
Memory SystemsFoundation Models
Research Hugging Face Papers Jul 29

Wonder: Video World Model Done Better

By Jiacong Xu, Hanwen Jiang, Zhixin Shu, Kalyan Sunkavalli, Vishal M. Patel, Yiqun Mei

88 score
AI Analysis

Wonder presents a video world model for real-time, camera-controllable world exploration using dense coordinate fields and efficient memory retrieval.

We present Wonder, a general-purpose video world model for real-time, camera-controllable world exploration. Given an image or a conditional video, Wonder constructs a playable world where users can navigate interactively by moving the camera, discovering unseen regions, and revisiting previously observed areas in real time and over a long-term horizon. Achieving this capability requires a system-level co-design of control method, memory mechanism, and training strategy. We introduce a novel cam
Generative ModelsVideo Generation
Research AlphaXiv Trending Jul 29

TurboVLA: Real-Time Vision-Language-Action Model at 32 Hz on an RTX 4090 with <1 GB VRAM

By Hengyi Xie, Chenfei Yao, Xianjin Wu, Xuanyang Xi, Yiping Tang, Di Xu, Yingying Zhu, Dingkang Liang, Xiang Bai, Han Ding

88 score
AI Analysis

TURBOVLA achieves real-time robotic manipulation at 32 Hz with under 1 GB VRAM on an RTX 4090.

TURBOVLA introduces an efficient vision-language-action model that achieves real-time robotic manipulation at 32 Hz with under 1 GB VRAM on an RTX 4090, directly mapping visual and linguistic features to actions. This approach matches or surpasses the manipulation performance of larger, LLM-centric systems, significantly lowering hardware requirements for deployment.
RoboticsEfficiency
Research AlphaXiv Trending Jul 29

Visko Orbis 1.0: A Live Model for Real-Time Interactive Long Video Generation

By Xiangbo Gao, Siyuan Yang, Ping He, Mingyang Wu, Yuheng Wu, Yushen Zuo, Jiongze Yu, Ryan Cui, Hongyuan Hua, Devin Ma, Xiao Jin, Yubo Yuan, Qing Yin, Jie Yang, Zhengzhong Tu

88 score
AI Analysis

Visko Orbis 1.0 is a live streaming model enabling real-time interactive long-video generation with dynamic prompt switching.

We present Visko Orbis 1.0, a Live Model for real-time, interactive long-video generation. Users can change the prompt at any moment during generation, and the update becomes visible in real time. Visko Orbis 1.0 supports long-form text-to-video, image-to-video, and video continuation, with multilingual prompts and prompt switching while generation is in progress. A bounded multi-scale memory preserves subjects, scenes, and style across chunks, sustaining hour-scale rollouts without evident qual
Video GenerationReal-Time AI
Research AlphaXiv Trending Jul 29

From Passive Video to Editable Experience: Physically Grounded Experience Synthesis for Embodied Intelligence

By Jia Luo

88 score
AI Analysis

Pegasus bridges the embodiment gap by translating human manipulation videos into robot-executable plans via task, affordance, and constraint graphs.

The key bottleneck in embodied AI is not model architecture but data. Although billions of human manipulation videos exist online, robots cannot directly learn from them due to the embodiment gap between human morphology and robot hardware. We introduce Pegasus, a low-resource framework that bridges this gap by translating human demonstrations into robot-learnable data through structured knowledge transfer. Instead of relying on raw video prompts, Pegasus constructs a graph-based intermediate re
Embodied AIRobotics
Research LessWrong Jul 29

Frontier Lab Employee Open Letter Calls For Being Able to Pace the Frontier

By Zvi

88 score
AI Analysis

Discusses an open letter signed by over 1,200 frontier lab employees calling for the ability to pace AI capability development to ensure adequate safety measures and oversight.

The most important open letter in years dropped yesterday. This letter noticeably increases my hope that we will manage to not die, and that we will otherwise be able to secure for ourselves a positive future, both by its impact and by the evidence it provides that such a letter can get this level of support. Signed by 1,224 employees of frontier labs including many heavy hitters, and now endorsed by both OpenAI and Anthropic, here is its full text, which I also endorse: AI could help create a d
AI GovernanceSafety
Research Hugging Face Papers Jul 29

MODUS: Decoder-Only Any-to-Any Modeling of Diverse Modalities

By Mingqiao Ye, Zhaochong An, Zhitong Gao, Xian Liu, François Fleuret, Chuan Li, Amir Zadeh, Serge Belongie, Afshin Dehghan, Jesse Allardice, David Mizrahi, Oğuzhan Fatih Kar, Roman Bachmann, Amir Zamir

43 score
AI Analysis

As first reported in Research yesterday, Modus investigates decoder-only any-to-any multimodal modeling, treating all modalities symmetrically without specialized heads or pipelines.

Any-to-any models predict any modality from any combination of others within a single network, a formulation used in multimodal vision and vision-language models, and increasingly in scientific domains such as ecology and astronomy. Existing any-to-any models are typically trained from scratch using encoder-decoder or diffusion architectures, impacting their performance and preventing them from using strong pre-trained decoder-only models as a prior. In this work, we investigate decoder-only any
Multimodal AIModel Architecture
Research AlphaXiv Trending Jul 29

Enfold: Folding World-Generator Computation into Predictive Representations for Efficient Embodied Control

By Weili Zeng, Yitong Xing, Fulong Liu, Chengqun Yang, Antao Xiang, Feng Tian, Jingnan Gao, Jisong Cai, Xin Wang, Xiaomin Wu, Yao Mu, Yichao Yan

87 score
AI Analysis

Enfold internalizes world-generator computation into predictive representations inferred directly from current visual context and language instructions.

World generative models are typically used through what they produce: a rendered future, a video-conditioned action, or latent context computed by a costly generative branch. We argue that their more reusable asset is the computation that constructs a future. As a generator transforms a corrupted future into a coherent trajectory, its intermediate states organize appearance, spatial layout, and interaction across levels of abstraction. Can this future-generative computation be internalized in a
World ModelsEmbodied AI
Research AlphaXiv Trending Jul 29

Practice Makes Policies: Bootstrapping and Consolidating Robotic Capabilities from Zero Human Demonstrations

By Jialiang Li, Yuhan Wang, Haojun Li, Gaojing Zhang, Yangtian Ye, Qipeng Liu, Haotian Liang, Wenzhao Lian

87 score
AI Analysis

HERO bootstraps and consolidates robotic manipulation capabilities from zero human demonstrations via VLM-guided heuristic reasoning.

HERO, a hierarchical agentic framework, enables robots to autonomously acquire and refine manipulation skills without initial human demonstrations by progressively evolving capabilities from VLM-guided heuristic reasoning to efficient, closed-loop visuomotor policies. The system successfully collected 664 forward and reverse subtask episodes, achieving an 86.0% average task success rate on diverse real-world manipulation tasks.
RoboticsEmbodied AI
Research Hugging Face Papers Jul 29

Mage-VL: An Efficient Codec-Native Streaming Multimodal Foundation Model

By Senqiao Yang, Kaichen Zhang, Zhaoyang Jia, Jinghao Guo, Yifei Shen, Xinjie Zhang, Xiaoyi Zhang, Haoqing Wang, Xiao Li, Peng Zhang, Xiang An, Yin Xie, Zhening Liu, Xun Guo, Jiahao Li, Shicheng Zheng, Jinglu Wang, Zongyu Guo, Wenxuan Xie, Zihan Zheng, Yuxuan Luo, Bin Li, Yan Lu

86 score
AI Analysis

Mage-VL is an efficient codec-native streaming foundation model using motion vectors to reduce visual token consumption by over 75%.

Standard vision-language models (VLMs) suffer from Moravec's paradox: they excel at complex offline visual reasoning but struggle with simple streaming perception tasks and process them inefficiently. We present Mage-VL, an efficient codec-native streaming foundation model for real-time multimodal understanding and interaction. At its core, our custom tokenizer, Mage-ViT, replaces uniform frame sampling by selectively encoding dynamic, entropy-rich regions using motion vectors and residual energ
Multimodal AIEfficiency
Research AlphaXiv Trending Jul 29

AngelSpec: Towards Real-World High Performance Inference with Speculative Decoding

By Hong Liu, Rui Cen, Junhan Shi, Guangshuo Qin, Jiebin Zhang, Tianyu Liu, Runzhi Fan, Guoliang Zhao, Ruobing Xie, Kai Zhang, Song Liu, Guanghua Yu, Jianchen Zhu

86 score
AI Analysis

AngelSpec optimizes LLM speculative decoding using workload-specialized drafter models and adaptive resource management (D-cut).

AngelSpec is a comprehensive framework that optimizes large language model inference through speculative decoding by employing workload-specialized drafter models and an adaptive runtime resource management strategy called D-cut. It achieves 1.98x to 2.40x speedup over autoregressive decoding and a 15.7% higher throughput compared to existing speculative methods at high concurrency, while largely preserving generation quality.
Inference OptimizationLarge Language Models