L4
PulseAugur coverage of L4 — every cluster mentioning L4 across labs, papers, and developer communities, ranked by signal.
7 day(s) with sentiment data
-
Ninebot builds 15,000-GPU cluster for multimodal AI in autonomous driving
Ninebot has established the industry's first L4-level cluster with nearly 15,000 GPUs, enabling its APEX multimodal foundation model to advance towards 100 billion parameters. This significant computing power upgrade al…
-
Leapmotor accelerates intelligent driving tech, targets top-tier by 2026
Leapmotor is accelerating its development of intelligent driving systems, aiming to be among the top contenders by 2026. The company has validated its world model technology route with its "Intelligent Driving 4.0" syst…
-
LLM inference optimization research details cost-quality-latency trade-offs · 2 sources tracked
Two new research papers explore the trade-offs between inference optimization techniques for large language models (LLMs), focusing on cost, quality, and latency. The first paper, "The Inference Engineering Pareto Atlas…
-
Pony.ai unveils electric T9 autonomous truck for production this year
Pony.ai has unveiled its fully electric T9 model, slated for production later this year. The company aims to commercialize its L4 autonomous driving technology by significantly reducing component costs and standardizing…
-
Neolix begins L4 autonomous vehicle testing in Japan
Neolix, a commercial autonomous delivery vehicle operator, has commenced L4-level autonomous driving tests in a closed environment at the Tokyo Distribution Center (TRC). This marks the company's first L4 test in Japan,…
-
WeRide, Uber secure Spain's first L4 robotaxi license for Madrid launch
WeRide, Uber, and AVOMO have secured Spain's first L4 autonomous vehicle operating license for passenger cars, issued by the Directorate-General for Traffic. This landmark approval allows for the deployment of robotaxis…
-
AI intelligence framework places GPT-4o, Claude 3.5 in collaborator tier
A theoretical framework called "Carbon-Silicon Dao" proposes a five-tier system for classifying intelligence, from basic pattern matching to self-aware consciousness. The framework uses a ten-dimensional scale to positi…
-
AI framework ARNAI enhances spinal image analysis by removing implant artifacts
Researchers have developed a novel AI framework called ARNAI, designed to improve the accuracy of spinal image segmentation and measurement, particularly in postoperative radiographs containing spinal implants. This fra…
-
Didi Autonomous Driving launches next-gen Robotaxi R2 for passenger testing
Didi Autonomous Driving has launched its next-generation Robotaxi, the R2, for passenger testing in select areas of Beijing and Guangzhou. This new vehicle, a collaboration with GAC Aion, features upgraded autonomous dr…
-
New AI system Anian developed for mental health support with safety gates
Researchers have developed Anian, a safety-gated multimodal AI backend designed for mental health support and mindfulness intervention routing, particularly for perinatal care. The system employs a hierarchical state re…
-
China releases mandatory national standard for autonomous driving safety
China has officially released a mandatory national standard for the safety requirements of intelligent connected vehicle autonomous driving systems. This standard, GB 44721—2026, will take effect on July 1, 2027, and ap…
-
Author addresses comments on MCP monitoring and security features
The author of this post is addressing comments and feedback received on previous articles regarding their work on MCP (likely a security or monitoring system). They clarify distinctions between different levels of monit…
-
Momenta secures Shenzhen Robotaxi testing permit, eyes 2027 mass production
Momenta has secured a road testing permit in Shenzhen, China, allowing for L4-level Robotaxi testing in the city. This development follows their recent announcement of a mass-produced custom Robotaxi project with SAIC a…
-
A.S. Watson Group IPO delayed; AI platform wins award at WAIC
A.S. Watson Group is reportedly considering delaying its planned IPO in London, originally slated for this fall, until next year. The dual listing in London and Hong Kong was expected to raise approximately $2 billion, …
-
Deep Principle AI Scientist Platform Wins SAIL Star Award at WAIC
The AI scientist platform developed by Deep Principle was honored with the SAIL Star award at the 2026 World Artificial Intelligence Conference (WAIC) in Shanghai. This platform integrates generative AI, first-principle…
-
格林美 plans share buyback; autonomous driving timeline revealed; DeepSeek eyes IPO
格林美 announced a plan to repurchase shares worth between 100 million and 160 million yuan. The repurchased shares are intended for equity incentives or employee stock ownership plans, with a maximum repurchase price of 1…
-
Huawei predicts L3/L4 autonomous driving scale by 2027-2028; Baidu pursues HK dual listing
Huawei executive Jin Yuzhi anticipates significant advancements in autonomous driving, with L3 commercialization expected by 2027 and L4 capabilities, including robotaxis and low-speed autonomous operations, to see pilo…
-
Tesla reverses course, distilling L4 Robotaxi models for L2 FSD
Tesla is shifting its autonomous driving development strategy, moving from an L2-to-L4 "bottom-up" approach to an L4-to-L2 "top-down" model. The company is now developing larger, more complex AI models for its L4-dedica…
-
New framework defines five levels of AI agentic autonomy
A new framework proposes five levels of agentic autonomy, ranging from basic assistance to fully managed orchestration. These levels are defined as L0: Assist, L1: Supervised action, L2: Scoped task delegation, L3: Goal…
-
Neural network toy model demonstrates computation in superposition
Researchers have developed a toy model to explore computation in superposition within neural networks. By training a single-hidden-layer ReLU network with 50 neurons to compute 100 sparse input features under an L4 loss…