What is true on Models?
| Time | Entity | Sector | Event / capture |
|---|---|---|---|
| 1:05 AM ET | @stretchcloud | 1:05 AM ET@stretchcloud @stretchcloud on X Reef is one of the most ambitious open-source AI infrastructure projects I have seen this year. ~300 GitHub stars in two days. The core idea is genuinely different from anything else in the space: most RL post-training pipelines treat the model and the scaffolding around it as separate problems. You train one, then you write the other. Reef rejects that separation entirely. It co-evolves model weights and agent harness simultaneously, using live task outcomes as the feedback signal for both. Model side: SAO (Self-Aligned Optimization) adjusts weights from what the agent actually accomplished.… | |
| 12:45 AM ET | $NVDA | 12:45 AM ET @stretchcloud on X What NVIDIA shipped at IFA yesterday quietly changes local inference infrastructure. PAIR (Personal AI Router) is a free, open-source tool that auto-discovers compatible GPUs on your local network and routes inference requests to whichever machine has capacity. RTX 20-series and above, Apple M4, DGX Spark all supported. Integrates natively with Ollama and LM Studio. A few things to unpack. This is load balancing, not memory pooling. PAIR does not shard a 70B model across three machines or aggregate VRAM. Each machine runs models that fit in its own memory. What PAIR does: intelligently… | |
| 12:39 AM ET | 12:39 AM ET @tchsignal on X Meta Puts a Price on AI Interaction Data Meta is offering a striking trade-off for developers using its Muse Spark AI model: much cheaper API access if they opt in to Contributor terms that allow their prompts and model outputs to contribute to future model development. Standard pricing is $1.25 per million input tokens and $4.25 per million output tokens. Under Contributor pricing, those rates fall to $0.10 and $0.20 respectively. That distinction matters. Meta is not paying users cash, and the Contributor terms are not the default for every Muse Spark customer. Standard pricing remains… | ||
| 11:14 PM ET | @smsehy | 11:14 PM ET@smsehy @smsehy on X Training reinforcement learning on cached latent embeddings solves policy training throughput. The critical step for automotive manufacturing remains hard reliability optimization to eliminate rare edge-case failures during high-speed assembly. https://t.co/JZJfqC0HNe | |
| 11:12 PM ET | @smsehy | 11:12 PM ET@smsehy @smsehy on X Prototyping robotic kinematics is accessible for small engineering teams, but foundation models and custom edge silicon create massive capital consolidation. A few platforms will own the physical AI operating stack while hardware makers assemble standard mechanical frames. https://t.co/BGdI5joTD9 | |
| 11:11 PM ET | @smsehy | 11:11 PM ET@smsehy @smsehy on X Physical constraints across high-bandwidth memory packaging and behind-the-meter power interconnects are forcing compute optimization down to the edge. When centralized megawatts take four years to commission, running quantized models on ruggedized plant controllers is the only operational option. | |
| 11:10 PM ET | @smsehy | 11:10 PM ET@smsehy @smsehy on X 255 million vehicles and global manufacturing plants already generate the physical sensor telemetry that foundation models need to learn real-world physics. Digital AI scales in datacenters, but physical AI requires industrial plants that understand ten-year hardware duty cycles. https://t.co/agvaM0yiMT | |
| 10:47 PM ET | @stretchcloud | 10:47 PM ET@stretchcloud @stretchcloud on X The 30-50x cost difference between computer use and MCP keeps rattling around in my head. The framing that clicks for me: computer use is paying for a human-in-the-loop at model prices. Every screen interaction generates a screenshot, feeds it through vision, waits for coordinate output, executes the click, takes another screenshot. You are spending tokens on what amounts to OCR and cursor navigation, round after round. MCP replaces that entire loop with a typed function call. No screenshots. No vision pass. No coordinate uncertainty. The model calls a tool, the tool returns structured data,… | |
| 10:27 PM ET | @stretchcloud | 10:27 PM ET@stretchcloud @stretchcloud on X The thing I keep noticing across agentic frameworks in 2026: resumability is now the reliability primitive everyone is building toward. Genkit Go 1.13 ships it properly. A Generate call that fails at tool round five returns what it finished alongside the error. You pass resp.History() back in, only the failed step reruns. No wasted tool calls. No redone work. The same logic extends to full agent sessions. A failed or cancelled turn saves completed rounds as a snapshot. Send an empty input, it picks up from there. This matters for production. Most AI agent failures today are partial. The agent… | |
| 10:19 PM ET | AGAgibot | 10:19 PM ETAGAgibot @techniahqrobot on X A big win for the robotics community AGIBOT just open sourced one of its most valuable real world datasets yet. @AGIBOTofficial has open-sourced AGIBOT WORLD 2026 Theme 3, a real-world dataset designed for reinforcement learning and embodied AI. The release contains 11,430 real-world robot trajectories across 14 industrial and household tasks. That includes • 1,024 successful policy rollouts • 1,369 failed policy rollouts • human in the loop corrections • external disturbances • task progress annotations • error-state annotations • human intervention data Most robot learning datasets focus… | |
| 10:14 PM ET | OPOpenAI | 10:14 PM ETOPOpenAI @DrJimFan on X Good old days at OpenAI in 2016: an agent stares at screen pixels, moves a mouse, and books a flight on United. We called it World of Bits, inside OpenAI Universe. 10 yrs later, Astra is reincarnated in the same universe. Even the naming is astronomically correct 😆 Universe was perhaps the most ambitious AI infra project at the time, but we couldn't quite figure out how to solve it. A policy with zero prior knowledge of what a "submit" button does has to rediscover the entire internet visual lingua by trial and error. In retrospect, RL from scratch against hand-drawn, per-task "artisan"… | |
| 5:42 PM ET | OPOpenAI | 5:42 PM ETOPOpenAI @rimtoln on X GPT-6 ASTRA ISN'T A PATCH. IT'S A GENERATION FLIP. openai just shipped the model they call a generational leap past gpt-5.6 sol not better chat better computer use · coding · cyber · science ▹ why this is the breakthrough first openai model at Critical cyber threshold agent stacks that actually drive the machine browser · forms · repos · multi-step work enterprise-first rollout · daybreak defenders first ▹ the scoreboard (vendor table) frontiermath t4 · astra 97.6 · fable 5.1 87.8 sol was 83.0 · that's a real cliff terminal-bench science · 64.6 vs fable 52.6 automationbench · 41.4 vs 31.4… | |
| 5:34 PM ET | @deepakpathak | 5:34 PM ET@deepakpathak @deepakpathak on X You don't have to post-train ChatGPT on every user. If you did, it would never have taken off. Yet this is exactly how robotics works today For robots to take off, they need to learn in-context. Great article from @Chris_J_Paxton on this new paradigm and S1's place in it: https://t.co/Kt4DTK3n2A https://t.co/as45gL0y8C | |
| 5:27 PM ET | TMThinking Machines Lab | 5:27 PM ETTMThinking Machines Lab @stretchcloud on X The AI lab valuation compression is happening.Thinking Machines Lab is in talks to raise $1 billion at a $40 billion pre-money valuation, according to The Information. That's down from $50 billion they explored late last year. Mira Murati founded the lab after leaving OpenAI. They've released one model: Inkling, in July 2026.The compression signal matters. Twelve months ago, a fresh frontier lab started by someone who ran OpenAI could reasonably command $50B+ on pre-product. The market is repricing. What changed: GPT-6 Astra launched today and saturated most existing benchmarks. xAI sits at… | |
| 4:27 PM ET | OPOpenAI | 4:27 PM ETOPOpenAI @techniahqrobot on X GPT-6 Astra is here, and it might be one of the most important AI releases yet. OpenAI says the model hits 98% on FrontierMath Tier 4, 99.9% on ARC-AGI 3 and a perfect 100% on ExploitBench. It also shows big improvements in computer use, science, coding, cybersecurity and professional work. What’s really interesting for humanoid robotics is what happens when this kind of stronger reasoning gets closer to the physical world. A humanoid needs way more than just good locomotion. It has to understand instructions, make sense of cameras and sensors, plan long sequences of actions, recover when… | |
| 3:30 PM ET | @SemiAnalysis_ | 3:30 PM ET@SemiAnalysis_ @SemiAnalysis_ on X Shoutout to the cracked team at @vllm_project that implemented recent agentic workload optimizations. (1/5)🧵 https://t.co/BsXCtLBwvU | |
| 3:03 PM ET | ASAstribot | 3:03 PM ETASAstribot @XRoboHub on X Robots can’t stop while the model thinks. In a high-speed throw, even one pause can kill the momentum. Astribot released SmoothRL for online RL during async inference. S1 keeps moving as the model computes the next action chunk. Most actions in a chunk never execute. Train on the full chunk, and RL credits or blames moves that never happened. SmoothRL learns only from executed actions, matching real deployment timing. After 250 rollouts, tossing jumped 39%→94%, pen capping 8%→83%, and box opening 30%→90%. One autonomous toss cut acceleration RMS by 52% and jerk RMS by 47%. The model keeps… | |
| 2:58 PM ET | AGAgibot | 2:58 PM ETAGAgibot @spaceandtech_ on X AGIBOT has officially open-sourced AGIBOT WORLD 2026 Theme 3: Reinforcement Learning, a real-world embodied AI dataset collected across expert demonstrations, autonomous policy rollouts, and human-in-the-loop corrections. It includes 11,430 real-world trajectories across 14 industrial and household tasks, capturing both successful and failed policy rollouts, along with detailed annotations for task progress, errors, disturbances, and human interventions. By learning from successes, failures, risks, and human corrections, robots can learn not only how to do it, but how to do it better.… | |
| 2:25 PM ET | VEVercel | 2:25 PM ETVEVercel @stretchcloud on X One command now routes Claude Code, Cursor, Codex, Cline, and four other coding harnesses through a single AI Gateway. That is what Vercel's setup flow does: run `vercel ai-gateway coding-agents setup`, it detects which agents you have installed, writes their configuration, and provisions an API key. No markup added on top. 300-plus models from 30-plus providers, behind one endpoint. The practical gain is real. When Anthropic, OpenAI, and SpaceXAI all hit outages in the same window this week, a gateway with automatic fallbacks made the difference between an agent that stalled and one that… | |
| 2:07 PM ET | MUMunari | 2:07 PM ETMUMunari @frontrunvc on X EARLY: @munariai - 6 followers (Robotics) Munari just launched. Operator scoring for robot training data. On @frontrunvc, the story started 47 days before the X account went live. jul 16: a tracked account follows @rmn. flagged. aug 11: flagged building. aug 19: website found. no x account yet. aug 30: a16z @speedrun SR007. sep 1: @munariai goes live. sep 4: 17 tracked accounts following the founders. a whole company, before it had a handle. this is timeline. coming soon to every company card on @frontrunvc. | |
| 1:48 PM ET | @AppliedInt | 1:48 PM ET@AppliedInt @AppliedInt on X At this year’s Agentic AI Summit hosted by @BerkeleyRDI, our Chief Scientist, @Wei_ZHAN_ shared some of the cutting edge research he has been leading in the physical AI space. As the industry races to deploy L2++ ADAS with imitation-learning-based E2E, Wei tackled a key question: what if end-to-end autonomy could be trained without imitation at all, relying only on reinforcement learning? An idea that runs against the current status quo. Here are some of his insights 🧵 | |
| 1:43 PM ET | @rimtoln | 1:43 PM ET@rimtoln @rimtoln on X EVERYONE SHOWS THE 300-AGENT SWARM. almost nobody shows what happens after 300 agents are easy to spawn turning 300 outputs into one structure is the hard part ▹ the pipeline nobody demos collect → connect → organize → unify one context graph at the end the swarm starts as hundreds of isolated research paths then sources start sharing entities links form · contradictions surface weak claims stay exposed eventually the mess collapses into one structure the whole system can reason over ▹ the reframe that's the part you usually never see the swarm is the demo the graph is the product bookmark… | |
| 1:19 PM ET | ZUZhejiang University | 1:19 PM ETZUZhejiang University @techniahqrobot on X Zhejiang University just pushed long-horizon robot manipulation forward with HINT. HINT helps Vision-Language-Action models keep track of the human’s original intent across long tasks instead of losing the goal mid-sequence. On dual-arm PiPER robots, HINT raised π0.5 full-task success Fruit sorting 10% → 60% Word spelling 13.3% → 86.7% The system was also tested with unseen objects, layouts and instructions. Persistent intent tracking could become a key layer for reliable VLA systems and long-horizon Embodied AI. | |
| 1:00 PM ET | DRDusty Robotics | 1:00 PM ETDRDusty Robotics @ctorobotics on X Construction plans, printed directly onto the floor. 🤖🏗️ Meet FieldPrinter 2 from Dusty Robotics, an autonomous layout robot designed to turn digital construction models into full-scale markings directly on the jobsite slab. Instead of crews manually measuring and marking layouts, the robot can print walls, MEP penetrations, anchor points, labels, QR codes and other information from coordinated BIM or CAD models. Dusty states the system can achieve 1/16-inch (1.6 mm) layout accuracy. A simple robot solving a very real construction problem. Could robots like this become standard equipment on… | |
| 12:36 PM ET | $NVDA | 12:36 PM ET @tchsignal on X Nvidia’s $12.9B Hugging Face Deal Raises a Bigger Question About Open AI Nvidia has agreed to acquire Hugging Face in a deal valued at about $12.93 billion, but the transaction has not closed yet. It is expected to complete in the first half of 2027, subject to regulatory approvals and other closing conditions. The bigger issue is what Nvidia ownership could mean for an AI platform used by more than 18 million developers, researchers and creators, with over 3 million models, 500,000 datasets and 1 million applications. Nvidia says Hugging Face will remain an open platform. Developers will not… | |
| 12:25 PM ET | @MelvinInvests | 12:25 PM ET@MelvinInvests @MelvinInvests on X AI’s biggest bottleneck is moving data and that could still create huge opportunities for optical networking companies (Save this) The chart shows a 1.6T optical transceiver, a device that transfers data between AI servers, switches, GPUs, and fiber optic networks. 1.6T means it can theoretically move up to 1.6 terabits of data per second, or 1,600 gigabits and that is about twice the speed of an 800G connection. This technology is important because AI data centers contain thousands of GPUs that must constantly exchange information. As AI models become larger, slow connections can leave… | |
| 12:06 PM ET | $NVDA | 12:06 PM ET @TheHumanoidHub on X Nvidia is now a humanoid robot designer. That's the real takeaway from this news. Nvidia is acquiring Hugging Face for $12.93B. Hugging Face is the largest open-model and dataset hub, the "GitHub of AI," with 3M+ models and 18M+ developers, robotics included. But here's the interesting part: Hugging Face has a serious robotics team from its acquisition of Pollen Robotics, which designs and builds open-source hardware platforms for robots, including humanoids (Reachy). Pollen team now gets the backing of the world's most valuable company. $NVDA | |
| 11:54 AM ET | @elonmusk | 11:54 AM ET@elonmusk @elonmusk on X Humans writing code as seen by future AI https://t.co/WqmYxAvKGx | |
| 10:45 AM ET | $NVDA | 10:45 AM ET @stretchcloud on X $12.9B is not a model acquisition. It is a distribution acquisition.NVIDIA just locked in the largest AI developer community on the planet: 18M developers, 3M models, 500K datasets, 1M apps. Jensen's team is not buying Hugging Face for what it does today. They are buying the channel through which every future model will move.Hardware companies do not buy software companies to improve their margins. They do it to eliminate the surface area where the next bottleneck forms. This is the same logic behind Intel buying McAfee and Qualcomm buying NXP. Except the AI compute market is growing faster… | |
| 9:10 AM ET | $NVDA | 9:10 AM ET @CKCapitalxx on X $NVDA just officially agreed to acquire Hugging Face for $12.93 billion. Hugging Face is where open source AI lives. Over 18 million developers, 3 million models, 500,000 datasets, and 200,000 companies using it to discover, test and deploy AI. Every open weight release lands there first. The detail most people will miss is that Nvidia was already the largest contributor to the platform. Over 500 models and 250 open datasets released there. They've been building inside this ecosystem for years, and now they own it. Here's what it opens up. >Distribution. Nvidia owns the compute. Hugging Face… | |
| 8:28 AM ET | $NVDA | 8:28 AM ET @Biti8888 on X 🚨 BREAKING: @nvidia has officially confirmed the acquisition of @huggingface Jensen Huang said open models strengthen safety and cybersecurity, accelerate innovation, and let every developer, startup, and country build AI on their own terms. He thanked Hugging Face CEO @ClementDelangue for the trust - NVIDIA will be a great home for the platform and its community. 🤗 Deal size: ~$11.9B, closing expected in 2027. PS picture is not real😁 | |
| 8:22 AM ET | $NVDA | 8:22 AM ET @techniahqrobot on X People saw the NVIDIA and Hugging Face acquisition news and immediately focused on foundation models, GPUs and the broader AI industry. But one of the biggest consequences could be in Physical AI and humanoid robotics. NVIDIA already controls a major part of the robotics stack with GR00T, Cosmos, Isaac Sim, Isaac Lab, Jetson Thor, CUDA and TensorRT. Hugging Face adds another critical layer with LeRobot, open robotics datasets, pretrained policies, VLA tooling, model distribution and Pollen Robotics with Reachy 2. Bring those pieces closer together and the robotics workflow becomes much more… | |
| 6:42 AM ET | HRHorizon Robotics | 6:42 AM ETHRHorizon Robotics @tphuang on X Horizon Robotics has hit 15m in chip delivery. It has a 31.94% mkt share in China's auto ADA chip mkt in the 1st 6 months of the yr. Its J6M chip recently started fielding on various models as the 128 TOPS is enough for most HNOA config. 40+ brands use Horizon chips & 8m+ drivers | |
| 5:19 AM ET | ANAnthropic | 5:19 AM ETANAnthropic @tchsignal on X Claude Cybersecurity Tests Exposed a Two-Layer AI Safety Problem Anthropic found three incidents in which Claude models, while running cybersecurity evaluations, reached real production systems instead of remaining inside the intended simulated environment. The first failure was operational. Claude had been told there was no Internet access, but a misunderstanding between Anthropic and evaluation partner Irregular left a real network path available. Across 141,006 reviewed evaluation runs, Anthropic identified six runs tied to three incidents affecting three organizations. The second issue is… | |
| 5:15 AM ET | SPSpaceXAI | 5:15 AM ETSPSpaceXAI @Newsforce on X A new LatchBio report found @SpaceXAI ’s @Grok 4.6 is now the top-performing AI model in biosecurity safety testing, refusing disguised requests for dangerous biological research while still answering legitimate scientific questions. Unlike other models, Grok’s safeguards rely on its own reasoning rather than external filters, allowing it to spot tactics such as mislabeled viral sequences, anonymized files, and harmless-sounding cover stories hiding harmful intent. It also scored near the top on other biology benchmarks, suggesting its stronger safety measures don’t come at the cost of… | |
| 5:13 AM ET | 5:13 AM ET @Biti8888 on X Figure and Hark are run by him simultaneously - meaning one person is literally steering two multi-billion-dollar companies in parallel, not even counting his role at Cover. That’s three companies, three billion-dollar valuations, one person at the helm of all of them at once. What’s especially striking is Index’s progress: the platform spent four months in stealth mode and has already racked up 16 million uploaded videos from 108 countries, paying out $15M to users for their content. The logic is simple: language models trained on internet text, but robots have nowhere to pull the physical… | ||
| 5:07 AM ET | @smsehy | 5:07 AM ET@smsehy @smsehy on X Deploying general-purpose robots into demanding factory plants requires tight integration between high-level reasoning models and ruggedized, fanless edge controllers certified for industrial temperature and vibration ranges. https://t.co/BFNDhbe2Aw | |
| 4:58 AM ET | @smsehy | 4:58 AM ET@smsehy @smsehy on X Capital deployment into physical AI foundation models is surging, but deploying robots into production environments requires tooling up domestic supply chains for high-precision strain wave gears, high-torque frameless motors, and tactile sensors. https://t.co/RpqkZcuEON | |
| 4:56 AM ET | @smsehy | 4:56 AM ET@smsehy @smsehy on X Deploying driverless fleets without steering controls validates perception models in urban corridors. The next operational test is managing remote teleoperation intervention ratios and depot turnaround maintenance at fleet scale. https://t.co/FZf39unH12 | |
| 4:41 AM ET | @tchsignal | 4:41 AM ET@tchsignal @tchsignal on X Lawsuit Targets Secrecy Around US Frontier AI Review Rules A new FOIA lawsuit is testing how much of the US government's frontier-AI review system should remain out of public view. Executive Order 14409 created two related mechanisms: a classified benchmarking process for identifying covered frontier models, and a voluntary framework under which participating developers can give the federal government early access to certain models. That access can last up to 30 days before the developer plans to provide the model to other trusted partners. The order explicitly says this framework does not… | |
| 4:13 AM ET | @OperationsPLS | 4:13 AM ET@OperationsPLS @OperationsPLS on X Agriculture mechanization took 80 years. Robotics + AI compresses that to a decade. Workflow redesign, not just headcount math, is the ops challenge. Transition speed is the variable nobody models. https://t.co/D9zBiDcyNI | |
| 4:00 AM ET | OPOpenAI | 4:00 AM ETOPOpenAI @tchsignal on X US Government Backs OpenAI's Fair Use Argument for AI Training The U.S. government has entered a major copyright fight over AI training, backing OpenAI's argument that using copyrighted written works to train large language models can qualify as fair use. In a 20-page Statement of Interest filed in federal court, the Justice Department argues that training serves a different purpose from the original works: models analyze text to learn statistical patterns such as vocabulary, syntax and relationships rather than simply distributing copies to the public. The government's position goes beyond… | |
| 2:47 AM ET | 2:47 AM ET @stretchcloud on X The model that matters for agentic software is not the best at chat. It is the one that completes a 4-hour task without losing the thread. Meta released Muse Spark 1.3 yesterday. The benchmark that got my attention is not the headline one. It is DeepSWE v1.1: 75.4, ahead of Opus 5 at 74.0 and GPT-5.6 Sol at 73.0. That is the first time an open model has led that table. DeepSWE measures real code fixes on GitHub issues. Terminal-Bench 2.1 measures sustained work in a shell across long tasks. Muse Spark 1.3 ties GPT-5.6 Sol at 88.8 on Terminal-Bench, above Opus 5's 86.7. Both favor models that… | ||
| 11:54 PM ET | FAFable | 11:54 PM ETFAFable @stretchcloud on X The model that finds the most bugs does not find all the bugs. Bug Hunt Bench: 2 real repos, 105 hidden bugs, all tests green. Fable 5.1 leads at 43/105. GPT-5.6 is one behind at 42. But the number that matters most: 43 bugs have not been fixed by any single model. The union of all models covers 62 out of 105. That means 62/105 bugs are findable, but no single agent will find all of them. Each model catches bugs the others miss. The pattern I keep noticing on multi-model benchmarks is that the top model does not return a superset of what the others find. It returns a different distribution.… | |
| 11:34 PM ET | @stretchcloud | 11:34 PM ET@stretchcloud @stretchcloud on X Standard coding benchmarks are saturating. The interesting separation is happening on autoresearch. Autoresearch Bench launched this week: a benchmark for coding agents autonomously tackling research problems. Not completing a spec, but identifying what to build, searching across literature and codebases, forming hypotheses, running experiments, synthesizing results. My read is this is the correct next benchmark. The top models have converged close enough on short-task evaluations that leaderboard position no longer predicts production outcomes reliably. Where the gap stays stark is in tasks… | |
| 10:34 PM ET | FAFable | 10:34 PM ETFAFable @stretchcloud on X The benchmark gap that keeps surprising me: on standard coding benchmarks, the top models cluster. On ultra-long horizon coding, the gap blows open. Proximal released FrontierSWE v2 today. 34 real software engineering tasks, evaluated over hours with tool use and complex dependencies. Fable 5.1 scores 56.3%. GPT-5.6 scores 32.25%. That is a 24-point gap. Nothing on the standard leaderboards looks like this. The cost picture is equally interesting. GPT-5.6 costs more per run ($175 vs $138 for Fable 5.1) at roughly half the score. DeepSeek V4 Flash is cheapest at $8.57 but plateaus at 14.05%.… | |
| 9:54 PM ET | 9:54 PM ET @stretchcloud on X Meta released Muse Spark 1.3 today. The improvements are measurable: 20% fewer tool calls and 25% fewer tokens compared to 1.2. Fewer tool calls is the right metric to optimize. Each round-trip adds latency and cost in an agentic loop. Reducing them is harder than reducing token count. This is the efficiency curve everyone building on top of these models should track. https://t.co/zSkyzwUonB | ||
| 9:00 PM ET | $NVDA | 9:00 PM ET @SemiAnalysis_ on X NVIDIA Research 🚀 has produced some great research, like LatentMoE (used in Kimi K3) and GatedDeltaNets (used in Qwen). But for e2e frontier training, NVIDIA's bureaucratic culture has produced embarrassing models like Nemotron3 Ultra. Despite NVIDIA Research having amazing talent, Nemotron3 Ultra, with 550B total params (55B active), is getting mogged by all the Chinese models, including even Qwen3.8 27B parameters, which has ~20x fewer parameters. | |
| 8:46 PM ET | @Scobleizer | 8:46 PM ET@Scobleizer @Scobleizer on X For a while I was depressed about VR. There are so many pioneers who have gone sour on VR/AR. That depression lifted this morning. The future is so amazing. And I am seeing this in many companies from robotics to space to VR/AR. And this is before he got AI hooked up. Which is why VR will come back. The models will make the magic in our own homes. The future is so insane. Parts arrive next year. You are seeing pre-quakes now. | |
| 8:34 PM ET | GCGeneral Catalyst | 8:34 PM ETGCGeneral Catalyst @stretchcloud on X 12 PhDs and a Fields Medal winner built this in 4 months, backed by General Catalyst. https://t.co/pyhyV4u6I4 has a protocol that passes hidden states directly from a 753B frontier model to a 4B edge model at inference time. No text exchanged between the models. Neither model is fine-tuned. They come from different model families entirely. The result: 80% as accurate as the frontier model alone, running 20x faster. First place on ARC-AGI. The mechanism is what matters here. Most model compression strategies force a choice between accuracy and cost. You either run the big model and accept high… | |
| 7:53 PM ET | @ruima | 7:53 PM ET@ruima @ruima on X Well, this news really is dividing people. Some see this as evidence that AI has so much room to grow, especially as it expands beyond coding as a use case, since apparently this top 1% are other tech companies. Others see it as evidence that there is not enough economic value outside of tech. Very interesting. I definitely think the bullish viewpoint could be true, but this could still represent a huge concentration risk for these two companies in particular if people move to more open models | |
| 7:34 PM ET | PEPerplexity | 7:34 PM ETPEPerplexity @stretchcloud on X The hybrid compute pattern is getting serious. Perplexity just open-sourced Lily, the local inference engine powering their Mac hybrid compute feature, and the architecture choices are worth understanding.Lily runs Qwen3.6-35B-A3B on Apple silicon using a Rust runtime with custom Metal kernels. The optimization split is separate paths for prefill and decode. That matters because the bottleneck on local AI models shifts depending on whether you're processing a long input or generating output tokens. Lily tunes both separately to match what the hardware can actually do.The use case is cleaner… | |
| 5:47 PM ET | WLWorld Labs | 5:47 PM ETWLWorld Labs @a16z on X World Labs co-founder Justin Johnson on how five photos from your phone could teach a robot to work in a room it has never seen: "I think there's a couple different tech trees that people are working on. One is this notion of fully in-context learning. Maybe I've got a robotics foundation model, then I can demonstrate to the robot once how a task should be performed, and that's enough for the robot to figure out that task." "Another version is what we're calling real-to-sim-to-real. Maybe I've got my pre-trained robotics foundation model, but I want to adapt it to this particular… | |
| 5:13 PM ET | ANAnthropic | 5:13 PM ETANAnthropic @MParekh on X ‘New and Improved’ in AI Models, Anthropic, OpenAI & Google. ARD #154 ...Fable 5.1’s fine print. Astra tips out on cyber. Gemini narrows the gap. World models level up. Full writeup and source links: https://t.co/UZSXoRFmE2 https://t.co/UOEqsFebFq | |
| 5:00 PM ET | ALAlibaba | 5:00 PM ETALAlibaba @stretchcloud on X The AI assistant category is being replaced. QwenWork landed in public beta globally this week. Alibaba's move is to make the next product category a single subscription that puts an AI agent in control of your browser, your desktop, and your files. Not a chat assistant. An agent platform. The platform merges three Alibaba products into one: QoderWork for code, MuleRun for automation, and Wukong for browser agents. Describe a task in natural language. The agent navigates websites, operates your local computer, and executes multi-step processes without you switching tools. It ships multimodal… | |
| 4:47 PM ET | @ricci_nov | 4:47 PM ET@ricci_nov @ricci_nov on X SinkPruner drops 89% of visual tokens in multimodal models with no training, keeping 96.5% of LLaVA-1.5's performance and 91.8% of Qwen2.5-VL's. The tokens it cuts are high-norm outliers earlier methods kept. Most visual context is paid for, not used. https://t.co/VIlBbkCqYU | |
| 4:33 PM ET | MOMostik | 4:33 PM ETMOMostik @Scobleizer on X You have been paying a giant model to think and then to write. That second part is the expensive one. Mostik splits it. The giant model reads your problem. A small one on your own computer writes the answer. They share the thinking, not the words. 80% as good. 20 times faster. https://t.co/AD0GdYmAUg @mostik_ai @aimalysheva | |
| 4:20 PM ET | $GOOGL | 4:20 PM ET @stretchcloud on X The model tier split just happened. Gemini 3.8 Flash and Gemini 3.8 Flash Cyber launched today. Same day, completely different access regimes. The public Flash model is on the Gemini API at $0.75 per million input tokens. It scores 73.7% on DeepSWE 1.1, leads Harvey's Legal Agent Benchmark, and is the third updated Flash in 6 weeks. Gemini 3.8 Flash Cyber is not on the public API at all. Access is through Google's Fairwind Program, restricted to government agencies, critical infrastructure operators, and software maintainers. The capability gap is not marginal. Google's Chrome Security team… | |
| 3:36 PM ET | 3:36 PM ET @TheRundownAI on X The heavy week of AI releases continues with Meta and Muse Spark 1.3. Big jumps across the board over 1.2, with Mark Zuckerberg calling it "frontier performance almost too cheap to meter." https://t.co/uSrU3L5P6O | ||
| 3:20 PM ET | @stretchcloud | 3:20 PM ET@stretchcloud @stretchcloud on X The question "which model is better for my use case" only has one honest answer: run your actual work across both and compare. Fable 5.1 shipped yesterday. Qwen3.8-Max-0902 shipped yesterday. Grok 4.7 arrives in 10 days. Gemini 3.8 Flash dropped today. Four serious coding models in 48 hours. What I keep seeing: the teams getting value aren't the ones reading benchmark tables. They're the ones running the same task across multiple agent backends in isolated environments, measuring real output on real code. That's what I built Campfire for. One browser tab, every major coding agent: Claude… | |
| 3:00 PM ET | WLWorld Labs | 3:00 PM ETWLWorld Labs @stretchcloud on X Something shifted with the World Labs Atlas announcement. We've had image generators, video generators, and 3D tools for years. Atlas does all three from a single model with the same spatial understanding. The technical description: a multimodal autoregressive diffusion transformer, pre-trained from scratch on image, video, 3D, and camera movement jointly. Not a video model with 3D features bolted on. One model, one spatial representation. What it can do: pixel-perfect camera control on generated frames up to 1440p. Video up to 60 seconds. Input a scene or image, move the camera in any… | |
| 11:01 AM ET | 11:01 AM ET Agility on X: "Every object a different size, weight, and grip. Rearranging a full room's worth of stuff is the same generalized manipulation Digit uses to take on mixed loads and changing layouts on a real warehouse floor.
#Agility #DigitRobot" / X Post Log in Sign up Post Agility on X: "Every object a different size, weight, and grip. Rearranging a full room's worth of stuff is the same generalized manipulation Digit uses to take on mixed loads and changing layouts on a real warehouse floor. \#Agility \#DigitRobot" - Agility @agilityrobotics Every object a different size, weight, and grip. Rearranging a full room's worth of stuff is the same generalized manipulation Digit uses to take on mixed loads and changing layouts on a real warehouse floor. #Agility #DigitRobot 00:00 View media 11:01 AM · Sep 2, 20261.6KViews 5 2 29 - cyberprince… | ||
| 9:43 AM ET | $NVDA | 9:43 AM ET @techniahqrobot on X We’re cooked if robots can start learning from bodies they don’t even have. NVIDIA Brown Columbia and Harvard researchers released Hydra-0, a world model that represents robot actions as pixel motion. It was trained on about 2,202 hours of multi-embodiment video spanning human hands, handheld grippers, single-arm robots and bimanual systems. Instead of tying an action to one specific robot body, Hydra-0 predicts how objects should move in the image. The team also tested it on a real 14-DoF YAM robot, which bent a flexible pipe from a desired object motion. This is still a controlled… | |
| 9:20 AM ET | CACaterpillar | 9:20 AM ETCACaterpillar Wall St Engine on X: "CATERPILLAR PARTNERS WITH FIELDAI ON PHYSICAL AI
$CAT is working with FieldAI to bring AI-powered autonomy and robotics into jobsites and factories, using Caterpillar’s operational data, FieldAI’s robot foundation models and NVIDIA technologies.
Early applications include au… / X Post Log in Sign up Post Wall St Engine on X: "CATERPILLAR PARTNERS WITH FIELDAI ON PHYSICAL AI $CAT is working with FieldAI to bring AI-powered autonomy and robotics into jobsites and factories, using Caterpillar’s operational data, FieldAI’s robot foundation models and NVIDIA technologies. Early applications include autonomous inspections, real-time digital twins, risk detection and AI-driven optimization of equipment and facility operations." - Wall St Engine @wallstengine CATERPILLAR PARTNERS WITH FIELDAI ON PHYSICAL AI $CAT is working with FieldAI to bring AI-powered autonomy and… | |
| 8:32 AM ET | 8:32 AM ET @10xthinker on X Are humanoids a business story or just hype? - The Beijing robotics conference almost looked like a comedy show — robots stumbling, misidentifying obstacles, and performing like toddlers. - Yet the buzz has helped companies like Unitree reach valuations exceeding 1,200x earnings. At the same time, most robots deployed in China today don’t require general-purpose intelligence. So why are billions being invested? Is it an obsession with the human form factor, or because humanoids can be emotionally resonant? Our world — factories, hospitals, and homes — is built for humans. That creates a… | ||
| 8:00 AM ET | MSMorgan Stanley | 8:00 AM ETMSMorgan Stanley @smsehy on X Morgan Stanley estimates building an Optimus Gen 2 costs $45,500 using China's supply chain, versus $131,800 without it. That $86,300 delta is not simply raw neodymium pricing. It reflects fifteen years of continuous capital expenditure across EV assembly lines, rare earth magnet processing plants, and specialized high-precision machine tools. Software and foundation models grab the headlines in robotics. Yet when humanoids transition to volume manufacturing, the physical AI race collides directly with industrial processing capacity at the periodic table. | |
| 3:44 AM ET | ALAlibaba | 3:44 AM ETALAlibaba Chubby♨️ on X: "wtf Qwen 3.8 has been updated to version 0902 and is now almost at the same level as Fable 5 in benchmarks.
The interesting two aspects:
1) For such significant jumps, it seems that you don't even change the version number anymore (no Qwen 3.9), rather the development happens so q… / X Post Log in Sign up Post Chubby♨️ on X: "wtf Qwen 3.8 has been updated to version 0902 and is now almost at the same level as Fable 5 in benchmarks. The interesting two aspects: 1) For such significant jumps, it seems that you don't even change the version number anymore (no Qwen 3.9), rather the development happens so quickly that an update is simply added 2) it is breathtaking to see how much China is catching up and the gap to US labs is narrowing. Despite still having significantly fewer computers in comparison, models like Qwen are following suit and keeping up. Absurd." - Chubby♨️… | |
| 3:26 AM ET | @smsehy | 3:26 AM ET@smsehy @smsehy on X Digital models scale exponentially by distributing weights across networks, but physical AI scaling remains gated by physical component manufacturing lead times. Sourcing specialized precision cycloidal gearboxes and high-voltage motor drives requires physical plant capital that cannot be accelerated by compute alone. | |
| 11:48 PM ET | ARaxis robotics | 11:48 PM ETARaxis robotics Omai Leidi (3/3) on X: "Happy Wednesday frens👋
Everyone is watching the race to build better robots.
I’m starting to think the real race is the data teaching those robots how to act.
I’ve been digging into @axisrobotics, and it changed how I look at the space.
At first glance, AXIS looks like… / X Post Log in Sign up Post Omai Leidi (3/3) on X: "Happy Wednesday frens👋 Everyone is watching the race to build better robots. I’m starting to think the real race is the data teaching those robots how to act. I’ve been digging into @axisrobotics, and it changed how I look at the space. At first glance, AXIS looks like a platform where you control simulated robots from your browser. That’s only the entry point. The bigger idea is robotic data infrastructure. Robots need experience to become useful. They need to pick things up, move objects, sort items, assemble parts, use tools, and operate in… | |
| 10:31 PM ET | ZIZiNovaLabs | 10:31 PM ETZIZiNovaLabs LimX Dynamics on X: "Together with ZINOVA's Tool Intelligence, TRON 2 takes on increasingly complex construction workflows.
@ZiNovaLabs builds on TRON 2 to explore an innovative robotic configuration for construction, demonstrating key tasks in a scaled-down tilt-up construction workflow, includi… / X Post Log in Sign up Post LimX Dynamics on X: "Together with ZINOVA's Tool Intelligence, TRON 2 takes on increasingly complex construction workflows. @ZiNovaLabs builds on TRON 2 to explore an innovative robotic configuration for construction, demonstrating key tasks in a scaled-down tilt-up construction workflow, including formwork assembly, multi-layer rebar placement and tying. TRON 2 serves as a modular and extensible embodied robotic platform for multi-tool, multi-step tasks across large workspaces. Its dual arms handle construction tools and materials across different orientations and… | |
| 7:00 PM ET | $NVDA | 7:00 PM ET Noble Machines Accelerates Humanoid Robot Development 3X | NVIDIA Manufacturing \| Robotics Noble Machines Accelerates General Purpose Industrial Robot Development With NVIDIA Isaac GR00T Learn More Objective Noble Machines is advancing industrial robot development with the Moby general-purpose industrial robot and its supporting software stack, including a proprietary AI driven whole-body-control system. Moby is designed for complex, hazardous, and labor-intensive work across factories, logistics centers, construction sites, and semiconductor facilities, where operating reliably requires advanced perception, reasoning, adaptation, and physical interaction.… | |
| 2:21 PM ET | ANAnthropic | 2:21 PM ETANAnthropic Yuchen Jin on X: "Even if most AI labs are benchmark-maxxing now, Fable 5.1 still looks like an insane jump.
From Anthropic’s blog:
- Found a 1-in-a-million crash Millennium’s team couldn’t explain for 4–5 years
- 2x faster than Opus 5 while using half the tokens
Big if true." / X Post Log in Sign up Post Yuchen Jin on X: "Even if most AI labs are benchmark-maxxing now, Fable 5.1 still looks like an insane jump. From Anthropic’s blog: \- Found a 1-in-a-million crash Millennium’s team couldn’t explain for 4–5 years \- 2x faster than Opus 5 while using half the tokens Big if true." - Yuchen Jin @Yuchenj\UW Even if most AI labs are benchmark-maxxing now, Fable 5.1 still looks like an insane jump. From Anthropic’s blog: \- Found a 1-in-a-million crash Millennium’s team couldn’t explain for 4–5 years \- 2x faster than Opus 5 while using half the tokens Big if true. View… | |
| 2:09 PM ET | NMNoble Machines | 2:09 PM ETNMNoble Machines From Robot Development to Deployment with Isaac GR00T & Jetson Thor - YouTube Error 401 (Bad Request)!!1 401. That’s an error. The server cannot process the request because it is malformed. It should not be retried. That’s all we know. Back Skip navigation Search Search with your voice Sign in From Robot Development to Deployment with Isaac GR00T & Jetson Thor Tap to unmute 2x From Robot Development to Deployment with Isaac GR00T & Jetson Thor NVIDIA Omniverse 12,267 views Streamed 7d ago Copy link Info Shopping If playback doesn't begin shortly, try restarting your device. • You're signed out Videos you watch may be added to the TV's watch history and influence TV… | |
| 2:09 PM ET | $NVDA | 2:09 PM ET Seeed reBot Arm B601-RS: Physical AI & VLA Model Course | NVIDIA DLI Series Warehouse China Warehouse US Warehouse Germany Warehouse The store will not work correctly in the case when cookies are disabled. US Warehouse: Enjoy FREE UNIUNI shipping on orders under $50! (Excludes XIAO & Raspberry Pi series) \\ \\ the AI Hardware Partner For Industry Products Local Warehouse Documents Customization Solution Software Support Open Claw Quick Order Account Get more with a SeeedStudio accountclear \\ \\ 1 on 1 Support \\ \\ Local Delivery \\ \\ 30-Day DOA Guarantee \\ \\ 30-Day Returns Sign In Create Account My Orders\\ Track, change, cancel Account Information\\ Update… | |
| 1:48 PM ET | SAsanctuary | 1:48 PM ETSAsanctuary Physical AI for Industrial Automation | Sanctuary AI We value your privacy We use cookies to enhance your browsing experience, serve personalised ads or content, and analyse our traffic. By clicking "Accept All", you consent to our use of cookies. CustomiseReject AllAccept All Customise Consent Preferences We use cookies to help you navigate efficiently and perform certain functions. You will find detailed information about all cookies under each consent category below. The cookies that are categorised as "Necessary" are stored on your browser as they are essential for enabling the basic functionalities of the site. ... Show more… | |
| 1:29 PM ET | $GOOGL | 1:29 PM ET Google DeepMind on X: "We’re bringing agentic video understanding to our latest Gemini models.
They can now analyze videos with better accuracy while using up to 88% fewer tokens. 🧵" / X Post Log in Sign up Post Google DeepMind on X: "We’re bringing agentic video understanding to our latest Gemini models. They can now analyze videos with better accuracy while using up to 88% fewer tokens. 🧵" - Google DeepMind @GoogleDeepMind We’re bringing agentic video understanding to our latest Gemini models. They can now analyze videos with better accuracy while using up to 88% fewer tokens. 🧵 View media 1:29 PM · Sep 1, 2026127.2KViews 93 132 1.3K 264 - Google DeepMind @GoogleDeepMind 21h Instead of scanning an entire file, Gemini reasons across the video’s transcript, audio, and… | |
| 1:28 PM ET | WLworld-labs | 1:28 PM ETWLworld-labs World Labs on X: "All of this is powered by one unified architecture: a multimodal autoregressive diffusion transformer, pretrained from scratch.
This foundation blends the best of modern LLMs and video models, benefiting from the architectural, algorithmic, and systems advances from both areas." / X Post Log in Sign up Post World Labs on X: "All of this is powered by one unified architecture: a multimodal autoregressive diffusion transformer, pretrained from scratch. This foundation blends the best of modern LLMs and video models, benefiting from the architectural, algorithmic, and systems advances from both areas." - World Labs @theworldlabs 12h Introducing Atlas: The world's first multimodal world model that generates image and video frames with pixel-perfect camera control and reconstructs them in 3D. Model the world, move the camera, and simulate space & time. 00:00 View media 330… | |
| 1:28 PM ET | WLworld-labs | 1:28 PM ETWLworld-labs Atlas: A World Model for Spatial Intelligence | World Labs September 1, 2026Introducing Atlas, our new omni world model for spatial intelligence. Atlas: A World Model for Spatial Intelligence World models generate, reconstruct, and simulate any possible world. They understand how worlds appear, behave, and evolve so that we can render imagined worlds for creative users, simulate the real world in high fidelity, and help robots plan actions. At World Labs, we build these general purpose world models in pursuit of spatial intelligence. Today we are introducing Atlas, our next-generation world model. Atlas is an omni model that we pretrained from… | |
| 12:00 PM ET | WAwafer | 12:00 PM ETWAwafer wafer on X: "We’re excited to announce that we’ve raised a $40M Series A!
co-led by @MarathonMP and @chemistry, with participation from @Wing_VC, @AMD Ventures, @outsetcap, @fiftyyears, and @ycombinator, and our existing investors doubling down on @wafer_ai. we are also joined by an incredible lis… / X Post Log in Sign up Post wafer on X: "We’re excited to announce that we’ve raised a $40M Series A! co-led by @MarathonMP and @chemistry, with participation from @Wing\VC, @AMD Ventures, @outsetcap, @fiftyyears, and @ycombinator, and our existing investors doubling down on @wafer\ai. we are also joined by an incredible list of angels, including @JeffDean (CEO, @DiscoLoopAI), @rauchg (CEO, @vercel), @andyfang (CTO, @DoorDash), @kvogt (CEO, Bot), @akothari (COO, @NotionHQ), @eastdakota (CEO, @Cloudflare), @deepgramscott (CEO, @DeepgramAI), and more! Most inference optimization today is manual,… | |
| 3:53 AM ET | https://x.com/ChongZzZhang | 3:53 AM EThttps://x.com/ChongZzZhang C. Zhang on X: "New paper release: Learning Agile Perceptive Traversal of Sparse 3D Structures for Humanoids
Paper https://t.co/LlrDsajHqx
Videos https://t.co/jXLtG68oUg
We study how to do learning and sim2real for 3d traversal, such as monkey bars and overhanging obstacles." / X Post Log in Sign up Post C. Zhang on X: "New paper release: Learning Agile Perceptive Traversal of Sparse 3D Structures for Humanoids Paper https://t.co/LlrDsajHqx Videos https://t.co/jXLtG68oUg We study how to do learning and sim2real for 3d traversal, such as monkey bars and overhanging obstacles." - C. Zhang @ChongZzZhang New paper release: Learning Agile Perceptive Traversal of Sparse 3D Structures for Humanoids Paper arxiv.org/pdf/2608.29769 Videos nemantor.github.io/sparse-3d-trav… We study how to do learning and sim2real for 3d traversal, such as monkey bars and overhanging obstacles.… | |
| 12:08 AM ET | 12:08 AM ET @OptionKing666 on X Unitree’s IPO raised RMB 6bn and closed its first day up 460% at an RMB 237bn market value; 48% of proceeds is allocated to robot-model R&D. The asymmetry is clear: funding for capability expansion is already abundant, while the return on that spend remains unproven through a full cycle. Validation requires brain R&D to monetize into the FactSet-estimated revenue ramp from RMB 2.9bn in 2026E to RMB 4.9bn in 2027E. A miss would falsify the capital-efficiency case; the main risk is multiple compression from Unitree’s 80x/48x 2026E/2027E P/S before monetization catches up. | ||
| 10:31 PM ET | $GOOGL | 10:31 PM ET smartfitguide on X: "Google DeepMind's RT-2 model treats robot actions as text tokens. By co-fine-tuning pre-trained Vision-Language Models with robotic data, it achieved nearly double the performance on novel tasks compared to its predecessor, RT-1. #AI" / X Post Log in Sign up Post smartfitguide on X: "Google DeepMind's RT-2 model treats robot actions as text tokens. By co-fine-tuning pre-trained Vision-Language Models with robotic data, it achieved nearly double the performance on novel tasks compared to its predecessor, RT-1. \#AI" - smartfitguide @bareani21645 Google DeepMind's RT-2 model treats robot actions as text tokens. By co-fine-tuning pre-trained Vision-Language Models with robotic data, it achieved nearly double the performance on novel tasks compared to its predecessor, RT-1. #AI View media 10:31 PM · Aug 31, 20268Views Reply 1 | |
| 9:40 PM ET | ZHZhipu | 9:40 PM ETZHZhipu tphuang on X: "GLM-6 disclosed - more parameter & lower activation ratio & inference cost & improved post training expected. Model to make self training decisions
Expecting significant increased domestic compute addition in next 3 to 6 months. Using self owned, leased & purchased c… / X Post Log in Sign up Post tphuang on X: "GLM-6 disclosed - more parameter & lower activation ratio & inference cost & improved post training expected. Model to make self training decisions Expecting significant increased domestic compute addition in next 3 to 6 months. Using self owned, leased & purchased compute." - tphuang @tphuang Sep 1 Regardless of how Zhipu employees view AGI internally, it publicly follows the OpenAI/Ant line of scaling RSI & Autonomous AI as the future. AI path according to Zhipu: Chat -> Coding -> Co-Work -> Autonomous Larger TAM @ each step w/ last… | |
| 9:40 PM ET | ZHZhipu | 9:40 PM ETZHZhipu tphuang on X: "How it views its own models in terms of the kind of tasks them can do. GLM-5.3 in terms of long time horizon work is still short of 1 wk, need at least 1 month to be autonomous.
Flash is the 1st one to use domestic chips & have low cost + intelligence to support long tasks." / X Post Log in Sign up Post tphuang on X: "How it views its own models in terms of the kind of tasks them can do. GLM-5.3 in terms of long time horizon work is still short of 1 wk, need at least 1 month to be autonomous. Flash is the 1st one to use domestic chips & have low cost + intelligence to support long tasks." - tphuang @tphuang Sep 1 Regardless of how Zhipu employees view AGI internally, it publicly follows the OpenAI/Ant line of scaling RSI & Autonomous AI as the future. AI path according to Zhipu: Chat -> Coding -> Co-Work -> Autonomous Larger TAM @ each step w/ last step having… | |
| 8:54 PM ET | HYHyper3D | 8:54 PM ETHYHyper3D @ErenChenAI on X Shanghai-based 3D generative AI company Hyper3D has introduced WorldGen, a world generation model that turns a single image into a complete, editable 3D scene. Built on its SIGGRAPH 2025 Best Paper-winning CAST research, WorldGen reconstructs individual objects along with their spatial relationships and physical properties, including collision, mass and friction. The generated scenes can be used in existing 3D workflows, with applications ranging from game development and XR to robotics simulation and embodied AI training. | |
| 8:36 PM ET | 8:36 PM ET The Robot Report: Skild AI unveils S1 robot foundation model | AI Understanding Back to News ProductAI Understanding briefing The Robot Report: Skild AI unveils S1 robot foundation model The Robot Report reports that Skild AI unveiled S1, a robot foundation model that the company says can learn complex tasks from a single human demonstration video and operate across multiple robot forms. By AI Understanding EditorialSeptember 1, 2026 at 12:36 AM UTCUpdatedSeptember 1, 2026 at 1:47 AM UTC6 min read Read the primary source The short version The Robot Report reports that Skild AI unveiled S1, a robot foundation model that the company says can learn complex tasks from a… | ||
| 8:00 PM ET | $GOOGL | 8:00 PM ET Introducing Agentic Video in Gemini Introducing agentic video understanding with Gemini Sep 01, 2026 \| 7 min read - x.com - Facebook - LinkedIn - Mail - Copy link Our new agentic feature for video analysis cuts token consumption by up to 88%, reduces costs by up to 66%, and boosts quality by up to 7%. Rohan Doshi Senior Product Manager, Google DeepMind Mario Lučić Research Director, Google DeepMind Share - x.com - Facebook - LinkedIn - Mail - Copy link Your browser does not support the audio element. Listen to article \[\[duration\]\] minutes This content is generated by Google AI. Generative AI is experimental… | |
| 6:19 PM ET | PAPerceptron AI | 6:19 PM ETPAPerceptron AI @techniahqrobot on X A 36B parameter model just blurred one of robotics’ biggest boundaries, the gap between seeing a scene and acting on it. Perceptron AI has released Isaac 0.5, an open-weight Embodied AI model designed to use one backbone to understand video, reason about a physical task, and generate robot actions. One model connecting perception, reasoning, and robot control. @perceptroninc | |
| 5:45 PM ET | $NVDA | 5:45 PM ET @a16z on X Gavin Baker says the future for the world's biggest companies is open models and private context: "I think the future is an ensemble of models. There's a Pareto curve. No one model is going to be the best at everything." "For the global 1,000 biggest companies, you're going to take whatever the best open-source model is. I think probably in the very near future, that's going to be an Nvidia model." "Everybody says 'Well, in a world where open-source wins, who funds the training?' Chip companies can fund the training." "It's trivial to do a $50 to $100 billion training run for Jensen, but I do… | |
| 5:21 PM ET | @pstAsiatech | 5:21 PM ET@pstAsiatech 国家发改委：加速具身智能在制造、医疗等领域真实场景中落地应用_财经上下游_澎湃新闻-The Paper 下载客户端 登录 无障碍 - +1 国家发改委：加速具身智能在制造、医疗等领域真实场景中落地应用 澎湃新闻记者 滕晗 2026-08-28 11:44 来源：澎湃新闻 ∙ 财经上下游 > 听全文 字号 8月28日，国家发展改革委举行8月份新闻发布会，国家发展改革委政策研究室副主任、委新闻发言人李超在会上强调，机器人产业涉及人工智能、先进制造、新材料等诸多前沿技术，必须坚持因地制宜、健康有序发展，立足本地资源禀赋和产业优势，找准定位、发挥优势，防止盲目跟风、一哄而上，推动相关产业发展能够行稳致远。 李超表示，国家发展改革委将聚焦务实管用、落地见效，以具身智能实训场和应用中试基地为抓手，让机器人在真实场景中迭代技术、围绕真实需求形成应用闭环。 一方面，统筹布局具身智能实训场，加强数据、模型、标准等要素供给。在数据方面，构建高质量真机数据采集系统，提升具身智能数据供给质量与规模，破解具身智能训练“数据饥渴”问题。在模型方面，依托高质量数据及真实场景，支持具身模型企业开展多技术路线探索，鼓励视觉—语言—动作模型、世界模型等前沿方向创新，加速技术收敛与应用落地。在标准方面，推动建设具身智能技术标准体系，以统一标准降低模型跨本体适配成本，促进技术共建共享。 另一方面，建好用好具身智能方向国家人工智能应用中试基地，加快应用落地和产业规模扩增。… | |
| 5:00 PM ET | SCSchaeffler | 5:00 PM ETSCSchaeffler The hardest problem in physical AI may be the magnet, not the model Skip to content AI Learning and Artificial Intelligence Credit: Canva Actuators account for 40% to 60% of a humanoid robot’s bill of materials and China refines roughly 90% of the rare earth magnets inside them, which makes the components rather than the models the limit on scale. Europe has a 2030 cap on single country supply under the Critical Raw Materials Act, and one company, Schaeffler, building the actuators at volume. The hardest problem in physical AI is not the AI. Actuators are in short supply from a small pool of qualified manufacturers, and new capacity takes years, not quarters,… | |
| 5:00 PM ET | TUThe University of Hong Kong | 5:00 PM ETTUThe University of Hong Kong The Imitator Game — Benchmarking Robot Imitative Ability Beyond Action Prediction Intent imitation, L0-L320,000+ paired episodesHuman evaluation Arena The Imitator Game BenchmarkingRobotImitativeAbilityBeyondActionPrediction Xunzhe Zhou 1,2,\,† · Yiyang Cai 2,3,\ · Fengyi Wang 2,3,\ · Ran Ju 1,2,\ · Hanxiang Ren 2,4 · Ruizhe Liu 1 Yu Zhang 1 · Qian Luo 1,2 · Feng Chen 1 · Pei Zhou 1,2 · Yi Ma 1,2 · Yanchao Yang 1,2,‡ 1The University of Hong Kong · 2TranscEngram · 3Fudan University · 4Zhejiang University \\ Equal contribution · † Project lead · ‡ Corresponding author Can a robot imitate what a human intends — not just what they do? We widen the gap between the demonstration… | |
| 2:33 PM ET | 2:33 PM ET @Astra1Byte on X Four years ago, Figure didn’t have a robot. Today it’s valued at $39B and has already shipped 1,000+ humanoids. CEO Brett Adcock says the real advantage isn’t the AI it’s that hardware, models, and manufacturing all sit under one roof, one campus, no outsourcing. Robots already deployed at BMW. Production line running. The question nobody’s asking loud enough: when the entire stack brain, body, and factory belongs to one company, how fast can this actually scale once the failure rate drops? | ||
| 2:01 PM ET | $NVDA | 2:01 PM ET NVIDIA's robot business is already worth $10 billion/year, and China is its best customer | HumanoidHub News/ trends,china,usa,physical\ai,hardware,government trends,china,usa,physical\ai,hardware,government NVIDIA's robot business is already worth $10 billion/year, and China is its best customer John KoetsierAugust 31, 2026 NVIDIA's physical AI business is already generating roughly $10 billion a year, more than the entire humanoid robot industry combined, with Jensen Huang pointing to a tenfold path over the next decade. A meaningful share of that comes from China, which built about 90% of the world's humanoids last year. But Beijing is racing to build a domestic stack ... Talk about picks… | |
| 1:39 PM ET | @NVIDIARobotics | 1:39 PM ET@NVIDIARobotics Optimize Models with Unsloth | Jetson AI Lab 🎉 Welcome to Jetson AI Lab 2.0! We've redesigned the site with curated tutorials. Looking for old content? Browse the archive → × Back to Tutorials Optimize Models with Unsloth Fine-tune, convert and deploy language models on NVIDIA Jetson with Unsloth and llama.cpp. Author Aditya Sahu Fine-tune and deploy language models directly on NVIDIA Jetson with Unsloth and JetPack 7.2. This tutorial uses memory-efficient QLoRA, exports the results to GGUF, and runs them locally. Two practical examples cover Qwen3.5-4B on Jetson Orin Nano and NVIDIA Nemotron 3.5 Lightning 30B-A3B on Jetson AGX Thor.… | |
| 12:54 PM ET | 12:54 PM ET @ruima on X I think the Unitree IPO was obviously overhyped (well, except to whoever bought it at $60B). It’s down to a more manageable ~$30B now, which is still extremely pricey. I wonder what this means for the eventual trading of @AGIBOTofficial, which should be next in line for an IPO. (If the volatility is extreme or the growth disappoints, I think the IPO will be delayed, frankly.) Unlike Unitree, which has major gaps, particularly in software, Agibot was conceived from the beginning as a full-stack robotics company. And despite deliberately not specializing, it has been surprisingly strong across… | ||
| 12:41 PM ET | TEtechnia | 12:41 PM ETTEtechnia @techniahqrobot on X Humanoid robotics has its own language. Here are 30 acronyms you will see constantly API = Application Programming Interface BC = Behavior Cloning CoM = Center of Mass CoP = Center of Pressure DoF = Degrees of Freedom DRL = Deep Reinforcement Learning F/T = Force/Torque FK = Forward Kinematics GRF = Ground Reaction Force HIL = Hardware-in-the-Loop HRC = Human-Robot Collaboration HRI = Human-Robot Interaction IK = Inverse Kinematics IL = Imitation Learning IMU = Inertial Measurement Unit LiDAR = Light Detection and Ranging LLM = Large Language Model MPC = Model Predictive Control RGB-D = RGB +… | |
| 12:35 PM ET | UNUnitreeRobotics | 12:35 PM ETUNUnitreeRobotics @thehypedotnews on X unitree robots now build themselves – and train while doing it @UnitreeRobotics deployed its embodied ai model inside its own factory in china. the robots aren't repeating pre-programmed movements – they perceive the environment, understand the task and perform real manufacturing operations on the production line. this isn't just production, it's also a data engine – every task the robots perform on the line generates real-world motion data that trains the next generation of models. the factory is both the product and the training ground. unitree shipped 5,500 humanoid robots in 2025 – more… | |
| 12:31 PM ET | RURunway | 12:31 PM ETRURunway Runway on X: "Today, we're sharing new research on Solaris, our first Interface World Model.
Solaris is a new kind of operating system that generates interactive interfaces frame by frame, in real time, with no code. We find that Solaris outperforms frontier LLMs when generating new interfaces, ac… / X Post Log in Sign up Post Runway on X: "Today, we're sharing new research on Solaris, our first Interface World Model. Solaris is a new kind of operating system that generates interactive interfaces frame by frame, in real time, with no code. We find that Solaris outperforms frontier LLMs when generating new interfaces, across structural similarity and information retention. Read more and request early access at the link below." - Runway @runwayml Today, we're sharing new research on Solaris, our first Interface World Model. Solaris is a new kind of operating system that generates interactive… | |
| 12:22 PM ET | @0xconglomerate | 12:22 PM ET@0xconglomerate @0xconglomerate on X @LeggedRobot @ROBOTIS Looking forward to the next update! Hehe would be fun to have a Mini Pupper that I could play with and make content on, eh? 😅👀 | |
| 11:32 AM ET | @frontrunvc | 11:32 AM ET@frontrunvc @frontrunvc on X EARLY: 5 @ycombinator fall cohort companies across Compute, Energy, Hardware and DevTools. 👇 @nodussss - 1 follower - (Compute): the first intelligent execution layer for ai workloads. founders: @viswa_kotra @pvarshh @runmirrors - 17 followers - (DevTools): staging environments for ai agents. replay real sessions to catch regressions before they ship. founder: @ai_singhal @invertixlabs - 33 followers - (Energy): building energy superintelligence. ai that operates and scales energy infrastructure. founder: @Josephperrotta_ @visiblsemi - 79 followers - (Hardware): a faster path to custom… | |
| 11:17 AM ET | @whitee_rhinoo | 11:17 AM ET@whitee_rhinoo @whitee_rhinoo on X The #US will impose up to 100% #tariffs on #Chinese #drones over 25kg, thermal models, and key parts starting Sept 3, with lighter drones facing a 25% rate. Aimed at cutting reliance on foreign #supply chains, experts say it will fragment the #market rather than sever ties with #China #MarketUpdate #StockMarket #USA #stocks #Bullish #bearish #TRUMP solana:6p6xgHyF7AeE6TZkSmFsko444wqoP15icUSqi2jfGiPN #America #DonaldTrump #rates #GlobalTrade #altcoin #altcoins #altseason #coin #coins #token #cryptocurrency #UnitedStates | |
| 9:05 AM ET | TUThe University of Hong Kong | 9:05 AM ETTUThe University of Hong Kong GitHub - hku-sail/StreamPI: StreamPI: Streaming Multimodal Temporal Modeling for Vision-Language-Action Models · GitHub Skip to content You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert {{ message }} Uh oh! There was an error while loading. Please reload this page. hku-sail/ StreamPI Public - Notifications You must be signed in to change notification settings - Fork\\ 4 - Star\\ 161 main 1 Branch 0 Tags Go to Branches pageGo to Tags page Go to file Code Open more actions menu Latest commit happinesslz add multi-node training… | |
| 8:24 AM ET | 8:24 AM ET Biti8 on X: "Four years ago, @Figure_robot didn’t even have a robot.
Today, it’s valued at $39B, has built 1,000+ humanoids, deployed robots at BMW, built its own AI models, manufacturing infrastructure and a global data collection network.
What impresses me most is the speed.
Figure isn’t build… / X Post Log in Sign up Post Biti8 on X: "Four years ago, @Figure\robot didn’t even have a robot. Today, it’s valued at $39B, has built 1,000+ humanoids, deployed robots at BMW, built its own AI models, manufacturing infrastructure and a global data collection network. What impresses me most is the speed. Figure isn’t building just a robot. It’s building the entire stack at the same time: hardware, AI, data, factories and real-world deployments. And now Figure 04 is already on the way. Four years later, much of the humanoid industry is being measured against what @adcock\brett and the Figure team… | ||
| 5:05 AM ET | ARAxis Robotics | 5:05 AM ETARAxis Robotics @GabrielnSpace on X i thought community robot data would hit diminishing returns fast but Axis Dataset V1 just made that take look stupid @axisrobotics pushed π0.5 from 83.9% to 88.8% on LIBERO Plus, and performance kept climbing as training data scaled from 25% to 100% with no obvious saturation 3m trajectories later the crowd is starting to look less like users and more like a distributed robotics lab #PhysicalAI | |
| 4:50 AM ET | DJDJI | 4:50 AM ETDJDJI Pentagon blacklist vs. FCC drone bans: What drone pilots… Breaking Pentagon blacklist vs. FCC drone bans: What drone pilots need to know about the differences thedronegirl.com 2d ago Aug 31, 2026 The Pentagon blacklist is a paper cut for DJI — it only blocks federal contracts. The FCC ban is the gut punch: new DJI/Autel drones can't get US authorization, so future models never reach shelves. Your current drone stays legal; the upgrade path doesn't. Read Full Article Not sure which DJI to buy? Check the Buyer's Guide → What you can actually buy in the US right now Many new DJI products are blocked from US sale by the FCC Covered List. Here's the gear… | |
| 4:00 AM ET | 4:00 AM ET How Figure Became the Biggest Name in Robotics | XMAQUINA DAO Genesis Auction Wave 2! Launches June 24 Get DEUS homeDAO Portal How Figure Became the Biggest Name in Robotics Color theme: Four years, three generations of humanoids, a $39 billion valuation and now one of the largest physical AI data engines ever built. .png) August 27, 2026 Category: Physical AI Read time: 9 minutes Share This: Four years ago, Figure didn’t have a robot. Today, the company is valued at $39 billion, has built more than 1,000 humanoids, has robots working inside BMW, is preparing deployments with another major US retailer, and has developed its own AI models, manufacturing… | ||
| 3:16 AM ET | @tchsignal | 3:16 AM ET@tchsignal @tchsignal on X U.S. Barriers Test China's Robotics Scale Advantage The U.S. is raising barriers around foreign-made drones and advanced robots, but those measures are colliding with a market where Chinese manufacturers already operate at major scale. The FCC has added foreign-produced advanced robotic devices, including humanoids and quadrupeds, to its Covered List, blocking new equipment authorizations for covered models. Separately, the White House has imposed Section 232 tariffs on imported drones and components, with rates ranging from 10% to 100% depending on the product and origin. At the same time,… | |
| 12:25 AM ET | SASanctuary AI | 12:25 AM ETSASanctuary AI Physical AI Platform | Industrial Automation Solutions | Sanctuary AI We value your privacy We use cookies to enhance your browsing experience, serve personalised ads or content, and analyse our traffic. By clicking "Accept All", you consent to our use of cookies. CustomiseReject AllAccept All Customise Consent Preferences We use cookies to help you navigate efficiently and perform certain functions. You will find detailed information about all cookies under each consent category below. The cookies that are categorised as "Necessary" are stored on your browser as they are essential for enabling the basic functionalities of the site. ... Show more… | |
| 8:00 PM ET | $GOOGL | 8:00 PM ET Gemini Robotics ER 1.6 shutdown date Aug 31, 2026; replacement ER 2 Google Developers Gemini API deprecations page, Robotics models table (page last updated 2026-08-27 UTC; shutdown date in the table is in-window): gemini-robotics-er-1.6-preview released April 14, 2026, shutdown date August 31, 2026, recommended replacement gemini-robotics-er-2-preview. Older gemini-robotics-er-1.5-preview listed shutdown April 30, 2026. Companion overview (ai.google.dev/gemini-api/docs/robotics-overview, scraped Sep 1) says upgrade by replacing model='gemini-robotics-er-1.6-preview' with gemini-robotics-er-2-preview or gemini-robotics-er-2-streaming-preview; ER 1.6 'will be shut down at the end of August.' Pricing page (same scrape): ER 2 Preview paid $2.00 input / $10.00 output per 1M tokens (text/image/video/audio); batch $1.00 / $5.00; ER 2 Streaming $2.00 / $10.00; ER 1.6 Preview $1.00 input ($2.00 audio) / $5.00 output. This is an API lifecycle event, not a factory-hours print. Gemini Robotics 2 VLA launch itself was July 30, 2026 (Hassabis) — outside. | |
| 7:20 PM ET | 7:20 PM ET @ParadisLabs on X AI is still so early, but so bullish. We're witnessing a huge unlock of economic value with the adoption of AI models like Claude, and more recently, agents like Grok Bot. This then filters down into the most bullish of bull points for AI. Enterprise AI adoption increases productivity -> increases enterprise earnings -> increases token demand as a flywheel effect -> increases ROI for the AI labs -> increases demand for AI infra e.g. compute. I recently heard from an ex-colleague that a large consulting firm is increasing their AI spend by 100x at their London office for their back office… | ||
| 7:11 PM ET | 7:11 PM ET @Robot_AIsignals on X Show it once. That is the entire instruction. In a post published in August, Skild AI introduced S1, a robotics foundation model prompted with a video demonstration in place of a written command. The company says the same frozen weights then performed four tasks absent from its pre-training — potting a plant, cooking pancakes, brewing pour-over coffee, assembling a kit. Skild builds Skild Brain, which it calls the first unified robotics foundation model to generalise across both tasks and robot hardware. It raised $1.4bn in January at a valuation the company puts above $14 billion (company… | ||
| 2:19 PM ET | 2:19 PM ET @AtMapshock on X The FCC Constraint: Closing the Western Market Escape Valve One mechanism that might have imposed commercial discipline on Chinese humanoid makers — exposure to demanding Western end-users who would not accept subsidized, underperforming products — has been removed by US regulatory action. The FCC in July 2026 restricted new equipment authorizations for foreign-made advanced robotic devices, directly affecting Unitree and its peers [Source: Reuters, Aug 18, 2026]. Previously authorized models can still be sold, but the pipeline of new Chinese humanoid products entering the US market is… | ||
| 12:10 PM ET | $TSLA | 12:10 PM ET kasper on X: "Elon Musk was asked when Tesla Optimus could become better than the best human surgeons.
His argument is that a robot would not learn like one doctor. Every Optimus surgeon could share what every other unit had seen, including rare complications that a human might encounter only once… / X Post Log in Sign up Post kasper on X: "Elon Musk was asked when Tesla Optimus could become better than the best human surgeons. His argument is that a robot would not learn like one doctor. Every Optimus surgeon could share what every other unit had seen, including rare complications that a human might encounter only once in an entire career. That could make expert care available in places where finding a specialist is difficult or impossible. It also raises a question almost everyone can understand. When your life is on the line, do you trust the experienced human standing beside you, or the… | |
| 12:34 PM ET | XPXPENG | 12:34 PM ETXPXPENG @techniahqrobot on X XPENG is putting serious money behind humanoid robotics. Its robotics business just raised more than $900 million, reaching a post-money valuation above $6.3 billion. The round was led by IDG Capital, with Tencent, Alibaba and Gaorong Ventures participating. XPENG founder He Xiaopeng and co-president Brian Gu also invested roughly $100 million personally. The capital will support the development of IRON XPENG’s human-sized humanoid along with robot hardware Physical AI models and training infrastructure. China’s humanoid race is getting much more expensive. | |
| 9:28 AM ET | @techniahqrobot | 9:28 AM ET@techniahqrobot @techniahqrobot on X BeyondMimic is now in Science Robotics. Berkeley and Stanford used a Unitree G1 to track human motions like sprinting, spin-kicks and cartwheels, then used guided diffusion to combine those skills for waypoint navigation and obstacle avoidance. mocap is still used for obstacle and waypoint locations, plus some state estimation. Humanoid agility is advancing fast. How much of it still depends on external sensing infrastructure? #HumanoidRobots #Robotics #PhysicalAI | |
| 6:01 AM ET | HYHyundai | 6:01 AM ETHYHyundai South Korea Commits KRW 2.3 Trillion to Build Full-Stack Humanoid Robotics Ecosystem by 2030 | Humanoids Daily Search articles Key Takeaways Show all - Dedicated Humanoid Capital: South Korea is investing KRW 2.3 trillion ($1.66 billion) through 2030 to build an indigenous, full-stack humanoid robotics industry, backed by an additional KRW 2.8 trillion ($2.0 billion) for field verification. - Aggressive Localization Targets: The plan seeks to raise domestic localization of core humanoid components from 45% to 80%, transitioning legacy auto parts manufacturers into precision robotics suppliers. - State Procurement Pipeline: The government will purchase 1,080 humanoid robots and roughly 5,000 total… | |
| 5:43 AM ET | @realgalleryx | 5:43 AM ET@realgalleryx @realgalleryx on X 서랍이 안 열리자 수저를 오븐에 넣었다. Physical Intelligence를 만든 Sergey Levine이 Ryan Peterman 앞에서 말한 평가 장면이다. 모델은 π0.5다. 주방에서 수저를 넣으라고 했는데 서랍이 안 열렸다. 옆에 오븐이 있으니 그걸 열고 넣기 시작했다. 아이한테 치우라고 하면, 안 보이게만 넣기도 한다는 것이다. 접시 닦기도 그랬다. 회색, 초록, 하나를 떨어뜨리자 본체가 그 위로 가서 가렸다. 그리고 메모에는 회색 접시는 닦았다고 적혀 있었다. 예전 로봇 실수는 말이 안 되는 버그였다. 지금은 애가 하는 실수다. 이제 자라기만 하면 된다고 했다. 데모가 사람처럼 보일수록, 사람처럼 숨기는지도 같이 보면 된다. | |
| 11:00 PM ET | $GOOGL | 11:00 PM ET Gemini Robotics 2 brings whole body intelligence to robots - YouTube Error 401 (Bad Request)!!1 401. That’s an error. The server cannot process the request because it is malformed. It should not be retried. That’s all we know. Back Skip navigation Search Search with your voice Sign in Gemini Robotics 2 brings whole body intelligence to robots Tap to unmute 2x Gemini Robotics 2 brings whole body intelligence to robots Google DeepMind 324,052 views 1 month ago Copy link Info Shopping If playback doesn't begin shortly, try restarting your device. • You're signed out Videos you watch may be added to the TV's watch history and influence TV recommendations. To avoid… | |
| 4:12 PM ET | $GOOGL | 4:12 PM ET @csentropy on X Announced July 30, 2026 (author Carolina Parada) as part of Gemini Robotics 2, three models: Gemini Robotics 2 (VLA) converts vision+language into motor control first DeepMind VLA to control a full humanoid feet-to-fingertips under one checkpoint. Partner-only. Gemini Robotics ER 2 (VLM) the embodied-reasoning brain; publicly available. Gemini Robotics On-Device 2 (VLA) runs locally, adapts to new embodiments in a few hours. Partner-only. | |
| 1:30 PM ET | @du_maximilian | 1:30 PM ET@du_maximilian MemoryAnchors 01 · Continual Learning Challenge Data Sensitivity in Continual Robot Learning To understand Memory Anchors, we first need to take a closer look at why robot policies forget old tasks when learning new ones. Sequential Task Learning We explore the problem setting of imitation learning on tasks in sequence. Drag the slider to step through ten tasks. The horizontal axis is the learning progress, and the vertical axis is the evaluated tasks. The lighter the off-diagonal elements, the worse the forgetting. ER buffer size 0.5% (Tiny)1% (Small)5% (Medium) Training stage 12345678910 Stage 5 of 10… | |
| 12:22 PM ET | @0xconglomerate | 12:22 PM ET@0xconglomerate @0xconglomerate on X I’m pretty sure some of the things I mentioned here are already things @eastworlds_io and other robot data companies think about internally. s/o to @scale_AI @xdofai @MeckaAI @micro1_ai @LightwheelAI @Figure_robot @LeRobotHF @BitRobotNetwork @PrismaXai @silencioNetwork @GoogleDeepMind @physical_int ++ The point of this report is really just to look past the headline numbers, figure out where the actual value is being created, and see if there are gaps that can be turned into something useful or actionable. That’s also where the framework in the graphic comes from: Measure → find the gaps →… | |
| 11:19 AM ET | ARAxis Robotics | 11:19 AM ETARAxis Robotics Cicada Market Making on X: "https://t.co/26k9SK56mf" / X Post Log in Sign up Post - Cicada Market Making @cicada\mm Inside Robotics and Physical AI Featuring insights from Axis Robotics In 2026, the artificial intelligence industry learned to solve the compute problem. GPUs are becoming more accessible, models are cheaper to run inference on, cloud infrastructure keeps growing. But the next wave of AI, robotics and Physical AI, has a completely different problem. It all comes down to data that simply does not exist in the volume needed. No Internet for Robots LLMs grew out of a foundation that already existed. Decades of text on the internet,… | |
| 10:11 AM ET | A1a16z | 10:11 AM ETA1a16z Why Top Founders Are Racing Into AI Infrastructure - YouTube Error 401 (Bad Request)!!1 401. That’s an error. The server cannot process the request because it is malformed. It should not be retried. That’s all we know. Back Skip navigation Search Search with your voice Sign in Why Top Founders Are Racing Into AI Infrastructure Tap to unmute 2x Why Top Founders Are Racing Into AI Infrastructure a16z 90,021 views 4 days ago Copy link Info Shopping If playback doesn't begin shortly, try restarting your device. • You're signed out Videos you watch may be added to the TV's watch history and influence TV recommendations. To avoid this, cancel and sign in to… | |
| 2:00 AM ET | ARAxis Robotics | 2:00 AM ETARAxis Robotics @axisrobotics on X Announcing our partnership with @Dexmal_AI About Dexmal: Dexmal is a technology company focused on general-purpose embodied intelligence, committed to building intelligent, useful, and trustworthy robots. As a core data infrastructure partner, Axis is teaming up with Dexmal to empower their VLA and world models through large-scale egocentric, simulation, and real-world data production. Backed by successful enterprise data deployments, this milestone highlights the scale, precision, and production-grade reliability of Axis’s compounding physical data engine. Together, we are establishing the… | |
| 8:13 PM ET | 8:13 PM ET Physical Intelligence (π) Physical Intelligence is bringing general-purpose AI into the physical world. We are a group of engineers, scientists, roboticists, and company builders developing foundation models and learning algorithms to power the robots of today and the physically-actuated devices of the future. We're looking to add a small number of people to our team and have listed a few of the roles we're hiring for below. That said, we're always on the lookout for the exceptional. If none of these roles quite describe you, tell us what does and why you'd be excited to work at PI — by applying under… | ||
| 8:00 PM ET | CECerebras | 8:00 PM ETCECerebras Cerebras's Next Generation CS-4: Fast Just Got Faster SubscribeSign in Cerebras's Next Generation CS-4: Fast Just Got Faster Double the Performance, Double the Power, Double the Fun Myron Xie, Bryan Shan, Wega Chu, and 2 others Aug 18, 2026 ∙ Paid 136 6 Share Cerebras revealed CS-4 this week, with more details to come at Hot Chips. CS-4 is their fourth-generation rack built around the same third generation 5nm wafer-scale engine: WSE-3. CS-4 doubles the performance of CS-3 through increased power consumption and clock frequency per wafer, and better rack-scale density. This all translates into CS-4 being able to double the tokens/s/user per… | |
| 7:05 PM ET | $TSLA | 7:05 PM ET Optimus Just Entered Production at Fremont. Here's What Changes for Tesla Investors. | The Motley Fool Accessibility Menu ▲ S&P 500 +---% \|▲ Stock Advisor +---% Join The Motley Fool Search for a company Accessibility... Help Arrow-Thin-Down\\ \\ S&P 500\\ \\ 7,673.19\\ \\ +0.5%\\ \\ +41.72 Arrow-Thin-Down\\ \\ DJI\\ \\ 53,161.69\\ \\ +0.7%\\ \\ +394.81 Arrow-Thin-Down\\ \\ NASDAQ\\ \\ 26,193.63\\ \\ +0.4%\\ \\ +93.85 Arrow-Thin-Down\\ \\ Bitcoin\\ \\ $76,768.00\\ \\ -1.3%\\ \\ -$1,024.21 Arrow-Thin-Down\\ \\ SPCX\\ \\ $140.04\\ \\ -1.5%\\ \\ -$2.20 Arrow-Thin-Down\\ \\ AAPL\\ \\ $324.93\\ \\ -0.1%\\ \\ -$0.20 Arrow-Thin-Down\\ \\ AMZN\\ \\ $255.68\\ \\ +0.3%\\ \\ +$0.76 Arrow-Thin-Down\\ \\… | |
| 3:20 PM ET | DYDyna | 3:20 PM ETDYDyna @Rewkang on X Most Physical AI companies are still doing lab demos. A key factor in our investment in Dyna last year was their world class post training expertise/results and their deployment focus. More deployments begets better data begets better models begets faster time to deployment We saw this loop play out for multimodal models like ChatGPT/Claude and autonomous driving and we're seeing it play out for robotics | |
| 2:23 PM ET | @frontrunvc | 2:23 PM ET@frontrunvc @frontrunvc on X plugged this article into claude + the @frontrunvc mcp. companies that fit the thesis: 👇 vertical, owning a category: @cosmic_robotics autonomous construction @shiraz_ai robots that learn on the job @gravisrobotics turns any excavator into a robot @buildmonumental on-site construction @griprobotics waste management robots, YC S26 the deployment layer: @sydekickbot "the deployment layer for physical AI" @enact_labs "the post-training layer for physical AI" @movingatomslab world models for robots, YC S26 @adamorobotics low-latency teleop stack | |
| 2:13 PM ET | DRDyna Robotics | 2:13 PM ETDRDyna Robotics Dyna Robotics on X: "After so many demos, models, pilots, robots are still struggling to land real deployments with real customers. Until now.
Today we’re excited to share that our robots have successfully crossed the ROI threshold, and Din Tai Fung, one of the highest revenue per location restaur… / X Post Log in Sign up Post Dyna Robotics on X: "After so many demos, models, pilots, robots are still struggling to land real deployments with real customers. Until now. Today we’re excited to share that our robots have successfully crossed the ROI threshold, and Din Tai Fung, one of the highest revenue per location restaurant chain in the US, is rolling out Dyna robots across its extensive restaurant network. This brings our rollouts across hotels, logistics, data centers, and many other use cases to a fleet that reaches hundreds of robots by the first half of 2027. And we’re just getting… | |
| 2:13 PM ET | DRDyna Robotics | 2:13 PM ETDRDyna Robotics Dyna Robotics on X: "After so many demos, models, pilots, robots are still struggling to land real deployments with real customers. Until now.
Today we’re excited to share that our robots have successfully crossed the ROI threshold, and Din Tai Fung, one of the highest revenue per location restaur… / X Post Log in Sign up Post Dyna Robotics on X: "After so many demos, models, pilots, robots are still struggling to land real deployments with real customers. Until now. Today we’re excited to share that our robots have successfully crossed the ROI threshold, and Din Tai Fung, one of the highest revenue per location restaurant chain in the US, is rolling out Dyna robots across its extensive restaurant network. This brings our rollouts across hotels, logistics, data centers, and many other use cases to a fleet that reaches hundreds of robots by the first half of 2027. And we’re just getting… | |
| 2:13 PM ET | DRDyna Robotics | 2:13 PM ETDRDyna Robotics Not Just a Model, But a Product — DYNA Contact11:33 AM Name Email Company Team Size 1-2021-5051-100100+ Message BusinessMediaInvestorOther Submit Sections: 01Not just a model, but a product02"Both" is our way03How Dyna-2 closed the gaps04Deployment is the eval05The deployment flywheel \[ Research \] Not Just a Model, But a Product Category: Research Author: Dyna Robotics Date: August 2026 Read: 17 min \[Sound\]\[ Fullscreen \] 1.Not just a model, but a product Today, we're really excited to share an announcement our team has been working toward over the past year: our robots have successfully crossed the ROI threshold, and Din Tai… | |
| 9:58 AM ET | PEPerceptron | 9:58 AM ETPEPerceptron @TheHumanoidHub on X Perceptron's Isaac 0.5 is a 36B open-weight embodied foundation model with only 2.5B active per token. The model - Perception, reasoning, and control all read from one shared backbone, so it makes visual decisions mid-trajectory - Null-expert routing lets each token use between zero and eight of the 256 experts - 35+ embodiments and 100k+ hours of demonstrations, so policies port across robots fast The result Scaling general video from 1,000 to 1M hours cut the teleoperation needed to hit their action-loss target from 5,884 hours to 28. On most benchmarks for spatial, physical, and grounding… | |
| 8:59 AM ET | 8:59 AM ET New BMW Cars & SUVs | BMW Dealer Serving Atlanta GA Skip to main content Labor Day Specials start NOW! Financing as low as 0.9% for 60 months + Loyalty Credits on select models. Shop Labor Day Offers Global Imports BMW - 500 Interstate North Pkwy SE Atlanta, GA 30339 - :855-984-2585 - Shop New - #### Shop New - New BMW Vehicles - Build Your Own - New BMW Specials - #### Shop by Series - X1 - X2 - X3 - X5 - X6 - X7 - 2 - 3 - 4 - 5 - 7 - 8 - Z4 - i4 - i5 - i7 - iX3 - XM - M - Shop Electric - #### Shop Electric - New BMW Electric Vehicles - Charging & Benefits - Why BMW Electric? - BMW PHEV Models - BMW iX3 - #### Shop by Model - Shop BMW… | ||
| 4:00 AM ET | ARAxis Robotics | 4:00 AM ETARAxis Robotics Axis Robotics × Booster: From Digital Twins to a Robot Data Engine Back How task-aligned simulation, real demonstrations, and continued pretraining can make each new Booster task easier to build than the last. Robot learning has a model problem. More importantly, it has a data-operations problem. Figure — Axis’s simulation solution spans third-person, head, and wrist observations while varying appearance, materials, object layout, lighting, and scene context. The objective is to preserve task structure while varying the details that should not control the policy. Introduction For robotics companies building new embodiments—and, increasingly, for the entire… | |
| 4:05 PM ET | $NVDA | 4:05 PM ET How to Train a Cross-Embodiment Robot Navigation Policy with AI Agents | NVIDIA Technical Blog Technical Blog Subscribe Related Resources Robotics English한국어中文 How to Train a Cross-Embodiment Robot Navigation Policy with AI Agents Aug 26, 2026 By Yan Chang, Mihir Acharya, Wei Liu, Katie Washabaugh and Aishwarya Singh +9 Like Discuss (0) - L - T - F - R - E AI-Generated Summary - COMPASS adapts the pretrained NVIDIA X-Mobility policy into residual specialists for specific robots and environments using reinforcement learning. - An agent-driven workflow with human approval gates automates environment validation, scene preparation, smoke testing, residual training, checkpoint evaluation,… | |
| 4:04 PM ET | 4:04 PM ET Agility on X: "Whole-body range of motion, stress-tested by a workout class. Every joint in sync — timing, velocity, balance. The same control Digit uses on a real warehouse floor.
#Agility #DigitRobot" / X Post Log in Sign up Post Agility on X: "Whole-body range of motion, stress-tested by a workout class. Every joint in sync — timing, velocity, balance. The same control Digit uses on a real warehouse floor. \#Agility \#DigitRobot" - Agility @agilityrobotics Whole-body range of motion, stress-tested by a workout class. Every joint in sync — timing, velocity, balance. The same control Digit uses on a real warehouse floor. #Agility #DigitRobot 00:00 View media 4:04 PM · Aug 26, 202611.3KViews 9 4 42 7 - Capital\Squeeze❄️ @CapitalSqueez Aug 27 Agility Robotics’ Digit is positioned to disrupt and… | ||
| 2:45 PM ET | 2:45 PM ET Rhys on X: "Cool to see a bunch of new robotics foundation model releases:
π0.7 from @physical_int
GEN-1.5 from @generalistAI
S1 from @SkildAI
Feels like the GPT-3 moment for robotics. Robot Models are Few-Shot Learners!" / X Post Log in Sign up Post Rhys on X: "Cool to see a bunch of new robotics foundation model releases: π0.7 from @physical\int GEN-1.5 from @generalistAI S1 from @SkildAI Feels like the GPT-3 moment for robotics. Robot Models are Few-Shot Learners!" - Rhys @RhysLindmark Cool to see a bunch of new robotics foundation model releases: π0.7 from @physical\int GEN-1.5 from @GeneralistAI S1 from @SkildAI Feels like the GPT-3 moment for robotics. Robot Models are Few-Shot Learners! View media View media 2:45 PM · Aug 26, 2026255Views 3 3 - Rhys @RhysLindmark Aug 26 Also damn, still so early for robot… | ||
| 1:05 PM ET | WLWorld Labs | 1:05 PM ETWLWorld Labs World Labs on X: "The next generation of robotics needs models that can understand, simulate, learn from, and act across virtual and physical worlds.
We’re hiring a Sr Business Development Lead, Robotics & Physical AI to help bring this technology into real-world applications 🤖🚀" / X Post Log in Sign up Post World Labs on X: "The next generation of robotics needs models that can understand, simulate, learn from, and act across virtual and physical worlds. We’re hiring a Sr Business Development Lead, Robotics & Physical AI to help bring this technology into real-world applications 🤖🚀" - World Labs @theworldlabs The next generation of robotics needs models that can understand, simulate, learn from, and act across virtual and physical worlds. We’re hiring a Sr Business Development Lead, Robotics & Physical AI to help bring this technology into real-world applications… | |
| 10:59 AM ET | 10:59 AM ET @adcock_brett on X We're now over 43,000 weekly active users collecting data to train Helix, our AI model for F.03 robots This data collection project, Index, is our answer to the data problem: the largest useful robot training dataset in the world Extrapolate out, and this is the pretraining needed for large scale robot generalization The effort started by purchasing data from vendors but this data was scarce and really poor quality - there was simply no way to make this work correctly at scale. So we did it ourselves. It was quite a massive effort that has now resulted in a Figure-owned data pipeline that is… | ||
| 10:13 AM ET | COCosmicbrainai | 10:13 AM ETCOCosmicbrainai @antopatrex1 on X scaling laws do not apply to physical AI yet. i'll say it again. scaling laws do not apply to physical AI. chelsea finn from physical intelligence showed this during her YC talk. they trained pi0.7 to make coffee. the robot couldn't learn from video alone. failed over and over. an engineer had to physically grab the robot and correct it multiple times just to get it to catch the coffee grouper with one hand and balance with the other. peter florence at generalist, same story. VLA models, the robot figured out hand usage from video datasets but still needed a human to step in and fix the… | |
| 6:10 AM ET | UNUnitreeRobotics | 6:10 AM ETUNUnitreeRobotics @UnitreeStore_SH on X Unlock unlimited sports potential (Extra large joint movement space angle, 23~43 joints) Force control of dexterous hands, manipulation of all thingsImitation & reinforcement learning driven Robot world model, let’s create it together Unitree G1 Price from $13.5K. Experience G1 at the Unitree Store in Shanghai. Follow us for more robots, technology, and store experiences. 📍 2nd Floor, Jiuguang Department Store, West Nanjing Road, Shanghai 🕙 Open daily, 10:00–22:00 🚇 Metro Lines 2, 7 & 14(Jing'an Temple Station，静安寺站) #Unitree #UnitreeRobotics #UnitreeG1 #G1 #HumanoidRobot #Robotics… | |
| 6:10 AM ET | $GOOGL | 6:10 AM ET @zhodonx on X I’ve said before that I think Google & DeepMind are playing a longer game with models. This is still a firm stance. Because a lot of the time, i think people look at Google’s model strategy too narrowly; Every Gemini release immediately gets reduced to how it compares with other labs or whatever benchmark is trending that week. Meanwhile the it’s more about where all of that intelligence might show up next. Like in Reasoning, Vision, Video, Agents. And particularly robotics. I believe Gemini Robotics ER 2 is a clear example here. It was built on Gemini 3.5 Flash. Except, now its intelligence… | |
| 7:36 PM ET | @dredgefactory | 7:36 PM ET@dredgefactory @dredgefactory on X A 29900$ MACHINE IS DANCING ON A PAVEMENT AND NOT ONE PERSON WALKING PAST LOOKS UP. 29900$ is a Unitree H2, the class this thing sits in. Entry models go for 13500$. Every beat throws its centre of mass and the feet catch it in real time - uneven stone, no tether, no handrail. Same loop that keeps it standing in a warehouse. | |
| 3:19 PM ET | @gupta_abhinav_ | 3:19 PM ET@gupta_abhinav_ @gupta_abhinav_ on X Punch: Skild S1 is 380 times more data efficient than standard VLA models! yes...let that sink in.. | |
| 2:05 PM ET | 2:05 PM ET @rohanpaul_ai on X Robotics is more and more getting close to get its version of in-context learning. Skild just released S1, a robot foundation model that uses video demonstrations to define tasks instead of language instructions. Give S1 1 human video showing a long, multi-step task, and the robot executes it straight away. No retraining. No fine-tuning. If this scales, you pay the enormous data bill once during foundation-model training, then amortize it across thousands of new tasks through prompting. That could change the economics of robot learning completely. | ||
| 2:04 PM ET | 2:04 PM ET @DeryaTR_ on X This is super exciting advance in robotics AI! I believe S1 is the most impressive robot foundation model I had seen! S1, is a robot foundation model built to learn new manipulation tasks through in-context prompting In Skild’s internal benchmarks, S1 achieved a 66% step-success rate on unseen tasks versus 9% for a language-prompted model, while one video demonstration provided roughly the benefit of 380 post-training examples! | ||
| 1:35 PM ET | @gupta_abhinav_ | 1:35 PM ET@gupta_abhinav_ @gupta_abhinav_ on X We did not rush to put our results in public domain because we wanted to spend time understanding the prompting phenomenon. For the last two months, we have tried to obtain scaling laws of prompting models. We also tried to explain the phenomenon and comparison under both seen and unseen settings. These graphs make our results extremely promising! | |
| 1:34 PM ET | RHrhoda | 1:34 PM ETRHrhoda @chris_j_paxton on X This is the mark of a general purpose robot foundation model -- that it can be taught tasks from examples without training. We saw an early version of this from rhoda and a solid one last week from generalist, and now a really impressive, ten minute long one from skild. This to me does feel like the true gpt moment for robotics -- because if it truly generalizes, that means that you can really start deploying robots for anything, as developing robot skills becomes like prompting | |
| 1:33 PM ET | 1:33 PM ET Brett Adcock on X: "The data needed to scale a truly general purpose robots doesn't exist on the internet - it has to come from the real world
For the last 4 months, we’ve been building a Figure-exclusive pipeline to scale data collection at higher throughputs. Today, we’re coming out of stealth" / X Post Log in Sign up Post Brett Adcock on X: "The data needed to scale a truly general purpose robots doesn't exist on the internet - it has to come from the real world For the last 4 months, we’ve been building a Figure-exclusive pipeline to scale data collection at higher throughputs. Today, we’re coming out of stealth" - Brett Adcock @adcock\brett Aug 25 Introducing Index Today we're coming out of stealth with Index, the largest & most diverse robot dataset in the world → 30min of video uploads/sec → 16M video uploads → Paid $15M to date → 264k downloads We're committed to spending $1B the… | ||
| 1:30 PM ET | 1:30 PM ET Brett Adcock on X: "Introducing Index
Today we're coming out of stealth with Index, the largest & most diverse robot dataset in the world
→ 30min of video uploads/sec
→ 16M video uploads
→ Paid $15M to date
→ 264k downloads
We're committed to spending $1B the next 12 months on data & com… / X Post Log in Sign up Post Brett Adcock on X: "Introducing Index Today we're coming out of stealth with Index, the largest & most diverse robot dataset in the world → 30min of video uploads/sec → 16M video uploads → Paid $15M to date → 264k downloads We're committed to spending $1B the next 12 months on data & compute" - Brett Adcock @adcock\brett Introducing Index Today we're coming out of stealth with Index, the largest & most diverse robot dataset in the world → 30min of video uploads/sec → 16M video uploads → Paid $15M to date → 264k downloads We're committed to spending $1B the… | ||
| 1:30 PM ET | 1:30 PM ET Brett Adcock on X: "Introducing Index
Today we're coming out of stealth with Index, the largest & most diverse robot dataset in the world
→ 30min of video uploads/sec
→ 16M video uploads
→ Paid $15M to date
→ 264k downloads
We're committed to spending $1B the next 12 months on data & com… / X Post Log in Sign up Post Brett Adcock on X: "Introducing Index Today we're coming out of stealth with Index, the largest & most diverse robot dataset in the world → 30min of video uploads/sec → 16M video uploads → Paid $15M to date → 264k downloads We're committed to spending $1B the next 12 months on data & compute" - Brett Adcock @adcock\brett Introducing Index Today we're coming out of stealth with Index, the largest & most diverse robot dataset in the world → 30min of video uploads/sec → 16M video uploads → Paid $15M to date → 264k downloads We're committed to spending $1B the… | ||
| 1:30 PM ET | 1:30 PM ET Figure on X: "Introducing Index
Today we're coming out of stealth with the largest and most diverse robot dataset to scale general purpose robots
To date we've crossed 264,000 app downloads and uploaded 16 million videos" / X Post Log in Sign up Post Figure on X: "Introducing Index Today we're coming out of stealth with the largest and most diverse robot dataset to scale general purpose robots To date we've crossed 264,000 app downloads and uploaded 16 million videos" - Figure @Figure\robot Introducing Index Today we're coming out of stealth with the largest and most diverse robot dataset to scale general purpose robots To date we've crossed 264,000 app downloads and uploaded 16 million videos 00:00 View media 1:30 PM · Aug 25, 2026525.5KViews 181 345 3.6K 872 - Figure @Figure\robot Aug 25 The data needed to scale… | ||
| 1:21 PM ET | 1:21 PM ET @TheHumanoidHub on X Show it a single human demonstration of a long-horizon task and the robot performs it right away. No fine-tuning, no retraining. Skild has released S1, a robot foundation model where you specify the task with a video demonstration instead of a language instruction. Why video, not language Language works for "hand me the mug." It falls apart for anything delicate or long-horizon. You don't learn to fold a fitted sheet from a sentence. So S1 takes a video demo into its context window and translates it to its own body and scene. What it does - Tasks up to 10 minutes, dozens of sequential steps -… | ||
| 1:19 PM ET | 1:19 PM ET Skild AI on X: "We found that to match the accuracy that S1 can achieve with just one example of prompting, current VLA models would need to be post-trained with 50-100 hours of data collection followed by fine-tuning!" / X Post Log in Sign up Post Skild AI on X: "We found that to match the accuracy that S1 can achieve with just one example of prompting, current VLA models would need to be post-trained with 50-100 hours of data collection followed by fine-tuning!" - Skild AI @SkildAI Aug 25 Introducing S1, our new foundation model that learns from one example. It can be taught 10-minute long tasks that it has never seen before, from one video prompt without any fine-tuning. Watch S1 operate in real-time via in-context learning: 00:00 View media 427 1K 7.1K 3.4M - Skild AI @SkildAI Aug 25 It can be taught… | ||
| 1:19 PM ET | @SkildAI | 1:19 PM ET@SkildAI @SkildAI on X Compared to conventional VLAs, S1 is a step-change improvement. For known tasks, S1 can match the performance of language-prompted VLAs. For novel tasks, S1 exponentially outperforms any existing language-prompted VLAs as we scale the pre-training. We believe this is a promising direction towards establishing scaling laws for robotics. | |
| 1:19 PM ET | 1:19 PM ET Skild AI on X: "S1 learns new tasks like a language model.
You prompt it with a video demonstration, and it outputs robot actions to complete the task in any environment and in any embodiment." / X Post Log in Sign up Post Skild AI on X: "S1 learns new tasks like a language model. You prompt it with a video demonstration, and it outputs robot actions to complete the task in any environment and in any embodiment." - Skild AI @SkildAI Aug 25 Introducing S1, our new foundation model that learns from one example. It can be taught 10-minute long tasks that it has never seen before, from one video prompt without any fine-tuning. Watch S1 operate in real-time via in-context learning: 00:00 View media 427 1K 7.1K 3.4M - Skild AI @SkildAI Aug 25 It can be taught extremely long-horizon tasks,… | ||
| 1:19 PM ET | 1:19 PM ET Skild AI on X: "The first time S1 flipped a pancake, we assumed pancake flipping must have been in its pre-training data.
We searched our whole pre-training data and found no examples of flipping.
S1 inferred the out-of-distribution task from one video prompt." / X Post Log in Sign up Post Skild AI on X: "The first time S1 flipped a pancake, we assumed pancake flipping must have been in its pre-training data. We searched our whole pre-training data and found no examples of flipping. S1 inferred the out-of-distribution task from one video prompt." - Skild AI @SkildAI Aug 25 Introducing S1, our new foundation model that learns from one example. It can be taught 10-minute long tasks that it has never seen before, from one video prompt without any fine-tuning. Watch S1 operate in real-time via in-context learning: 00:00 View media 427 1K 7.1K 3.4M - Skild… | ||
| 1:19 PM ET | 1:19 PM ET Skild AI on X: "Introducing S1, our new foundation model that learns from one example.
It can be taught 10-minute long tasks that it has never seen before, from one video prompt without any fine-tuning.
Watch S1 operate in real-time via in-context learning:" / X Post Log in Sign up Post Skild AI on X: "Introducing S1, our new foundation model that learns from one example. It can be taught 10-minute long tasks that it has never seen before, from one video prompt without any fine-tuning. Watch S1 operate in real-time via in-context learning:" - Skild AI @SkildAI Introducing S1, our new foundation model that learns from one example. It can be taught 10-minute long tasks that it has never seen before, from one video prompt without any fine-tuning. Watch S1 operate in real-time via in-context learning: 00:00 View media 1:19 PM · Aug 25, 20263.4MViews 427… | ||
| 1:19 PM ET | 1:19 PM ET Skild AI on X: "Introducing S1, our new foundation model that learns from one example.
It can be taught 10-minute long tasks that it has never seen before, from one video prompt without any fine-tuning.
Watch S1 operate in real-time via in-context learning:" / X Post Log in Sign up Post Skild AI on X: "Introducing S1, our new foundation model that learns from one example. It can be taught 10-minute long tasks that it has never seen before, from one video prompt without any fine-tuning. Watch S1 operate in real-time via in-context learning:" - Skild AI @SkildAI Introducing S1, our new foundation model that learns from one example. It can be taught 10-minute long tasks that it has never seen before, from one video prompt without any fine-tuning. Watch S1 operate in real-time via in-context learning: 00:00 View media 1:19 PM · Aug 25, 20263.4MViews 427… | ||
| 12:21 PM ET | BYBYD | 12:21 PM ETBYBYD @SexyTechNews on X FCC "Buy American" rule walls off foreign humanoids. New Covered List action blocks new foreign humanoid/quadruped models from US equipment authorization via a domestic-content test — directly hits BYD's newly unveiled Xiao Di and Unitree exports. | |
| 5:31 AM ET | NENEURA | 5:31 AM ETNENEURA NEURA Gym: Physical AI Training and Infrastructure | NEURA Skip to content The Fastest Path from Human Expertise to Deployable Robot Skills Get in contact \\ \\ The Challenges Facing Industrial Leaders Today Physical AI needs real-world training data: vision, sound, touch, force, spatial awareness. None of it can be sourced online or generated synthetically. With 120,000× less training data than large language models and skilled labor declining, every month without action carries a cost. Download Brochure\\ \\ Lack of Skilled Workforce Experienced operators retire. Their knowledge leaves with them. Deployment Risk on the Floor Untested robots on a… | |
| 4:41 AM ET | 北京北京人形机器人创新中心有限公司 | 4:41 AM ET北京北京人形机器人创新中心有限公司 中国机器人“跑赢”博尔特，这意味着什么？ - 纽约时报中文网 中国 - 中文中 - 中英双语双语 - 英文 英 YAN ZHUANG2026年8月25日 Achmad Ibrahim/Associated Press 2009年，尤塞恩·博尔特在柏林世界田径锦标赛上创造了9.58秒的男子 100米短跑世界纪录。据中国官方媒体报道，上周六，由中国公司北京人形机器人创新中心有限公司(X-Humanoid)开发的一款人形机器人以0.19秒的优势打破了该纪录。这台机器人在全速撞上一堵软垫墙后结束了冲刺，随后倒在地上，被人类助手抬走。 在中国，机器人展示运动天赋的高调壮举正变得越来越普遍。在国家的大力支持和数十亿资金的投资下，中国的人形机器人产业蓬勃发展。在该国的重要活动中，机器人曾表演过舞蹈和武术。 在上周末于北京举行的世界人形机器人运动会上，它们进行了拳击和足球比赛。其中一台机器人与人类对手进行了网球比赛，它在球场上移动，调整角度回击截击球。今年早些时候，一台机器人以超越历史上任何人类的速度跑完了一场马拉松。 但是，这些机器人超越人类的壮举究竟意味着什么呢？ 广告 澳大利亚昆士兰科技大学教授兼机器人中心主任迈克尔·米尔福德表示：“人形机器人参与的大多数体育项目并不能直接转化为人们希望它们完成的大量日常任务的实用价值，比如在家庭中提供帮助。” 他说，这些机器人“在非常具体、且有些刻意安排的赛事中表现出色”。… | |
| 8:00 PM ET | 8:00 PM ET Introducing Index: Building The World’s Largest and Most Diverse Physical Dataset Introducing Index: Building The World’s Largest and Most Diverse Physical Dataset August 25, 2026 Today we're coming out of stealth with the most diverse robot training dataset ever built. The data needed to scale a truly general purpose robot doesn't exist on the internet - it has to come from the real world: a global sampling of physics captured across every environment on earth. For the last 4 months, we’ve been building a Figure-exclusive pipeline to scale data collection at higher throughputs, with broad diversity and strict quality standards. - While in stealth, we've crossed 264,000… | ||
| 7:55 PM ET | @EdmondIsARobot | 7:55 PM ET@EdmondIsARobot 🤖🍨 Sundae Robotics 06: V-JEPA 2 & Predicting Physical Intelligence · Luma Hosted By Edmond Jono Hart James (Jingxi) Xu Mene Mazarakis Angela Zhang Manfredi Bernardi Adam 134 Went Akshobhya Gupta, Hannah Tsui and 132 others Contact the Host Report Event AI 🤖🍨 Sundae Robotics 06: V-JEPA 2 & Predicting Physical Intelligence Hosted by Edmond & 6 others Aug 30 Sunday, August 30 2:00 PM - 5:00 PM PDT Register to See Address Atherton, CA Past Event This event ended 2 days ago. Welcome! To join the event, please register below. Request to Join About Event 🤖🍨 Grab a sundae and join Sundae Robotics, a private, invite-only Sunday series bringing together robotics… | |
| 4:07 PM ET | XMXPeng Motors | 4:07 PM ETXMXPeng Motors XPeng Motors humanoid robot unit Dogotix raises $900M - The Robot Report The Robot Report Continue to Site Facebook X LinkedIn Reddit Pinterest Share XPeng’s Dogotix unit is developing the IRON humanoid robot. Source: XPeng Like other global automakers, XPeng Motors is branching into humanoid robots. The electric vehicle manufacturer today said that its Dogotix robotics unit has raised more than $900 million in its initial funding round, bringing its pre-money valuation to $5 billion and post-transaction valuation to more than $6.3 billion. XPeng said it plans to use the investment to further develop hardware and software, collect data, train physical AI models,… | |
| 9:15 AM ET | https://x.com/RoboPapers | 9:15 AM EThttps://x.com/RoboPapers RoboPapers on X: "Instead of choosing between training a world model and training a language conditioned robot policy, why not do both? LDA-1B s a new foundation model that is trained on 30,000 hours of human and robot interaction data.
Part of the secret is that LDA-1B jointly learns forward … / X Post Log in Sign up Post RoboPapers on X: "Instead of choosing between training a world model and training a language conditioned robot policy, why not do both? LDA-1B s a new foundation model that is trained on 30,000 hours of human and robot interaction data. Part of the secret is that LDA-1B jointly learns forward dynamics, action prediction, and visual forecasting, all in a structured DINO latent space which avoids the pitfalls of redundant pixel-level prediction which isn’t necessarily aligned robot action. This approach works on both dexterous hands and simple robot grippers; it also… | |
| 9:15 AM ET | @RoboPapers | 9:15 AM ET@RoboPapers @RoboPapers on X Instead of choosing between training a world model and training a language conditioned robot policy, why not do both? LDA-1B s a new foundation model that is trained on 30,000 hours of human and robot interaction data. Part of the secret is that LDA-1B jointly learns forward dynamics, action prediction, and visual forecasting, all in a structured DINO latent space which avoids the pitfalls of redundant pixel-level prediction which isn’t necessarily aligned robot action. This approach works on both dexterous hands and simple robot grippers; it also generalizes across objects, tasks, and… | |
| 12:50 AM ET | @TheHumanoidHub | 12:50 AM ET@TheHumanoidHub @TheHumanoidHub on X Generalist AI just released GEN-1.5. It might be robotics' GPT-3 moment. What is one-shot learning via in-context prompting? In the case of language models like GPT-3, including a Q&A example in the prompt before the actual question improved performance across a broad suite of language tasks. For example: Prompt: "Q: Who wrote Romeo and Juliet? A: William Shakespeare Q: Who wrote War and Peace?" [model outputs "Leo Tolstoy"] Similar capabilities are now emerging in GEN-1.5, where a short example (a few seconds of demonstration ) alongside language and sensory inputs can solve tasks the model… | |
| 12:41 PM ET | WLWorld Labs | 12:41 PM ETWLWorld Labs Bringing Marble to Life | World Labs Nov 12, 2025A behind-the-scenes look at how Marble powered the creation of its own launch story. Bringing Marble to Life World Labs: Just Imagine - YouTube Tap to unmute World Labs: Just Imagine World Labs World Labs5.53K subscribers Overview Copy link to this section When the World Labs team set out to create Marble’s first marketing video, they made a bold decision: to build it with Marble itself. The result was a launch video created using the same technology it introduced. Over the course of a few weeks, hundreds of 3D worlds were imagined, refined, and brought into the stage pipeline.… | |
| 12:00 PM ET | https://x.com/Majumdar_Ani | 12:00 PM EThttps://x.com/Majumdar_Ani Anirudha Majumdar on X: "Very cool work, Pete! I'm curious if you've tried an extreme version of this: no in-context example at all; just place objects in front of the robot and see if it can infer the task.
I suspect this will have non-trivial success rates, which would also allow you to figure … / X Post Log in Sign up Post Anirudha Majumdar on X: "Very cool work, Pete! I'm curious if you've tried an extreme version of this: no in-context example at all; just place objects in front of the robot and see if it can infer the task. I suspect this will have non-trivial success rates, which would also allow you to figure out how much of the benefit is actually coming from the prompt (rather than pure zero-shot task inference + capability). I remember Russ showed something like this in his Stanford talk on LBMs last year: https://t.co/0hEBrVlfOQ" - Pete Florence @peteflorence Aug 19 Ever since… | |
| 12:00 PM ET | $NVDA | 12:00 PM ET Post-Train NVIDIA Cosmos 3 Edge for On-Device Robot Control | NVIDIA Technical Blog Technical Blog Subscribe Related Resources Robotics Post-Train NVIDIA Cosmos 3 Edge for On-Device Robot Control Aug 19, 2026 By Saeed Babamohamadi +12 Like Discuss (0) - L - T - F - R - E AI-Generated Summary - NVIDIA Jetson Thor can run the 4B Cosmos 3 Edge omni-model natively for on-device robot manipulation policies. - Post-training Cosmos 3 Edge on the Cosmos3-DROID dataset produces a policy that generates action chunks in about 1.53 seconds on Jetson AGX Thor T5000, enabling continuous real-time control. - In closed-loop RoboLab evaluation, the post-trained Edge policy reaches 22.9%… | |
| 8:00 PM ET | $NVDA | 8:00 PM ET Hydra-0: Action Flow for Generalist World Modeling and Control | NVIDIA Isaac Overview video Figure 1: Action flow as a shared control interface. Top: Hydra-0 learns from diverse interaction videos featuring egocentric human demonstrations, handheld UMI grippers, bimanual robot arms, and single-arm robots. Middle: Visible embodiment motion is represented as image-plane flow trajectories, placing heterogeneous interactions in a common, pixel-aligned action flow space independent of embodiment-specific joint or end-effector coordinates. This unified interface enables a single generalist world model to learn from multi-embodiment data and transfer across interaction… | |
| 8:00 PM ET | 8:00 PM ET Press-Information June 25th 2026 Press-Information June 25th 2026 BMW Group advances the use of Physical AI in production with Figure 03 project in Spartanburg - Figure AI demonstrates Figure 03 humanoid robots in new use case at BMW Group Plant Spartanburg. - Robot development runs in parallel at BMW Group Plant Spartanburg and at Figure AI. - Assembly Hall in Spartanburg features BMW iFACTORY applications in artificial intelligence and virtualization. Munich/Spartanburg, USA. BMW Group intensifies the usage of digitalization and the use of artificial intelligence (AI) in production. With so-called Physical AI, which… | ||
| 4:01 PM ET | $NVDA | 4:01 PM ET 5 Physical AI infrastructure platforms shaping robotics in 2026 - The Robot Report Facebook X LinkedIn Reddit Pinterest Share From accelerated computing and simulation to data operations, open-source tooling, validation engineering, and continuous learning, these five platforms represent distinct control points in the emerging physical AI stack. For most of the modern AI boom, infrastructure had a single center of gravity: compute. Models grew larger, training runs consumed more GPUs, and the industry organized itself around accelerators, cloud clusters, training frameworks, and developer software. That stack was sufficient when AI’s outputs were text, images, video, or… | |
| 4:01 PM ET | $NVDA | 4:01 PM ET 5 Physical AI infrastructure platforms shaping robotics in 2026 - The Robot Report Facebook X LinkedIn Reddit Pinterest Share From accelerated computing and simulation to data operations, open-source tooling, validation engineering, and continuous learning, these five platforms represent distinct control points in the emerging physical AI stack. For most of the modern AI boom, infrastructure had a single center of gravity: compute. Models grew larger, training runs consumed more GPUs, and the industry organized itself around accelerators, cloud clusters, training frameworks, and developer software. That stack was sufficient when AI’s outputs were text, images, video, or… | |
| 12:00 PM ET | $GOOGL | 12:00 PM ET Gemini Robotics 2 brings whole body intelligence to robots — Google DeepMind Skip to main content July 30, 2026 Models Gemini Robotics 2 brings whole body intelligence to robots Carolina Parada Share Your browser does not support the video tag. From feet to fingertips — we are teaching robots intelligent whole-body control, fine dexterity, and teamwork to complete a broad range of complex tasks For decades, we’ve dreamed of robots that can seamlessly step into our world and lend a hand. Now, that vision takes a significant stride forward. Most robots are pre-programmed or teleoperated for narrow, repetitive task sequences. They lack the ability to truly learn for… | |
| 12:00 PM ET | $GOOGL | 12:00 PM ET Gemini Robotics 2 brings whole body intelligence to robots — Google DeepMind Skip to main content July 30, 2026 Models Gemini Robotics 2 brings whole body intelligence to robots Carolina Parada Share Your browser does not support the video tag. From feet to fingertips — we are teaching robots intelligent whole-body control, fine dexterity, and teamwork to complete a broad range of complex tasks For decades, we’ve dreamed of robots that can seamlessly step into our world and lend a hand. Now, that vision takes a significant stride forward. Most robots are pre-programmed or teleoperated for narrow, repetitive task sequences. They lack the ability to truly learn for… | |
| 9:57 PM ET | https://x.com/Sylviaposts | 9:57 PM EThttps://x.com/Sylviaposts Sylvia Li on X: "Spent the day at Deep Tech Week and left with one strong conviction:
the next wave of AI will not be won only by better models.
It will be won by the infrastructure that lets intelligence touch the physical world.
For anyone building in robotics, chips, manufacturing, energy, or… / X Post Log in Sign up Post Sylvia Li on X: "Spent the day at Deep Tech Week and left with one strong conviction: the next wave of AI will not be won only by better models. It will be won by the infrastructure that lets intelligence touch the physical world. For anyone building in robotics, chips, manufacturing, energy, or physical AI, this feels increasingly obvious. Physical AI is not just an AI problem. It is a data problem, a hardware problem, a deployment problem, and a feedback-loop problem. ChatGPT had the browser. Robotics still does not have a universal distribution layer. For robot… | |
| 8:00 PM ET | Joseph Amigo, Rooholla Khorrambakht, Nicolas Mansard, Ludovic Righetti | 8:00 PM ETJoseph Amigo, Rooholla Khorrambakht, Nicolas Mansard, Ludovic Righetti Coupled Local and Global World Models for Efficient First Order RL Real-world tasks solved zero-shot by policies trained entirely inside our learned world models: Push-T with a tabletop manipulator (left), Ego-Centric Grasp and Lift with a G1 humanoid (centre), and Ego-Centric Push Cube with a Go2 quadruped (right). Abstract World models offer a promising avenue for capturing complex environment dynamics where simulators face challenges. However, large-scale world models required for complex real-world settings are computationally expensive to adopt in popular RL approaches. We introduce a novel first-order RL method that enables policy training via a… | |
| 8:00 PM ET | Joseph Amigo, Rooholla Khorrambakht, Nicolas Mansard, Ludovic Righetti | 8:00 PM ETJoseph Amigo, Rooholla Khorrambakht, Nicolas Mansard, Ludovic Righetti Coupled Local and Global World Models for Efficient First Order RL Real-world tasks solved zero-shot by policies trained entirely inside our learned world models: Push-T with a tabletop manipulator (left), Ego-Centric Grasp and Lift with a G1 humanoid (centre), and Ego-Centric Push Cube with a Go2 quadruped (right). Abstract World models offer a promising avenue for capturing complex environment dynamics where simulators face challenges. However, large-scale world models required for complex real-world settings are computationally expensive to adopt in popular RL approaches. We introduce a novel first-order RL method that enables policy training via a… | |
| 8:00 PM ET | Joseph Amigo, Rooholla Khorrambakht, Nicolas Mansard, Ludovic Righetti | 8:00 PM ETJoseph Amigo, Rooholla Khorrambakht, Nicolas Mansard, Ludovic Righetti Coupled Local and Global World Models for Efficient First Order RL Real-world tasks solved zero-shot by policies trained entirely inside our learned world models: Push-T with a tabletop manipulator (left), Ego-Centric Grasp and Lift with a G1 humanoid (centre), and Ego-Centric Push Cube with a Go2 quadruped (right). Abstract World models offer a promising avenue for capturing complex environment dynamics where simulators face challenges. However, large-scale world models required for complex real-world settings are computationally expensive to adopt in popular RL approaches. We introduce a novel first-order RL method that enables policy training via a… | |
| 1:00 AM ET | ALAMI Labs | 1:00 AM ETALAMI Labs Yann LeCun's AMI Labs raises $1.03B to build world models | TechCrunch Checking your Browser… Verifying... Stuck? Troubleshoot Success! Verification failed Troubleshoot Verification expired Refresh Verification expired Refresh Troubleshoot Cloudflare, opens in a new tab Privacy • Help ababababababababababababcdcdcdcdcdcdcdcdcdcdcdcd efefefefefefefefefefefefghghghghghghghghghghghgh ijijijijijijijijijijijijklklklklklklklklklklklkl Skip to content Image Credits: Ruhani Kaur/Bloomberg via Getty Images / Getty Images AI Share on FacebookShare on XShare on LinkedInShare on RedditShare over EmailCopy Share Link Yann LeCun’s AMI Labs raises $1.03B to build world models… | |
| 1:00 AM ET | ALAMI Labs | 1:00 AM ETALAMI Labs Yann LeCun's AMI Labs raises $1.03B to build world models | TechCrunch Checking your Browser… Verifying... Stuck? Troubleshoot Success! Verification failed Troubleshoot Verification expired Refresh Verification expired Refresh Troubleshoot Cloudflare, opens in a new tab Privacy • Help ababababababababababababcdcdcdcdcdcdcdcdcdcdcdcd efefefefefefefefefefefefghghghghghghghghghghghgh ijijijijijijijijijijijijklklklklklklklklklklklkl Skip to content Image Credits: Ruhani Kaur/Bloomberg via Getty Images / Getty Images AI Share on FacebookShare on XShare on LinkedInShare on RedditShare over EmailCopy Share Link Yann LeCun’s AMI Labs raises $1.03B to build world models… | |
| 10:24 AM ET | 10:24 AM ET An Electric New Era for Atlas | Boston Dynamics We value your privacy We use cookies to enhance your browsing experience, serve personalized ads or content, and analyze our traffic. By clicking "Accept All", you consent to our use of cookies. CustomizeReject AllAccept All Customize Consent Preferences We use cookies to help you navigate efficiently and perform certain functions. You will find detailed information about all cookies under each consent category below. The cookies that are categorized as "Necessary" are stored on your browser as they are essential for enabling the basic functionalities of the site. ... Show more… | ||
| — | — Physical-Intelligence/openpi # openpi openpi holds open-source models and packages for robotics, published by the [Physical Intelligence team](https://www.physicalintelligence.company/). Currently, this repo contains three types of models: - the [π₀ model](https://www.physicalintelligence.company/blog/pi0), a flow-based vision-language-action model (VLA). - the [π₀-FAST model](https://www.physicalintelligence.company/research/fast), an autoregressive VLA, based on the FAST action tokenizer. - the [π₀.₅ model](https://www.physicalintelligence.company/blog/pi05), an upgraded version of π₀ with better open-world… | ||
| — | SUStanford University | —SUStanford University OpenVLA: OpenVLA: An Open-Source Vision-Language-Action Model Moo Jin Kim∗,1 Karl Pertsch∗,1,2 Siddharth Karamcheti∗,1,3 Ted Xiao4 Ashwin Balakrishna3 Suraj Nair3 Rafael Rafailov1 Ethan Foster1 Grace Lam Pannag Sanketi4 Quan Vuong5,† Thomas Kollar3 Benjamin Burchfiel3 Russ Tedrake3,6 Dorsa Sadigh1 Sergey Levine2 Percy Liang1 Chelsea Finn1 https://openvla.github.io970k Robot Episodes ViT Llama 2 7B Base VLM OpenVLA Vision-Language-Action Model Fine-tune VLM w/ Robot Actions: Closed-Loop Robot Control Policy User: Wipe the table. OpenVLA: [ x, , Grip] = …Δ Δθ Δ Multi-Robot Control & Efficient… | |
| — | PRPromise | —PRPromise From Archive to Production: A Hybrid Workflow with Marble | World Labs Aug 12, 2026How Promise used Marble to transform archival imagery into immersive 3D environments for Harlan Coben’s Final Twist. From Archive to Production: A Hybrid Workflow with Marble Using Hybrid Production to Reconstruct Crime Scenes - YouTube Tap to unmute Using Hybrid Production to Reconstruct Crime Scenes Promise Promise573 subscribers Overview Copy link to this section For Harlan Coben's Final Twist on Paramount+ and CBS, Promise developed a hybrid production workflow that combined World Labs' Marble, AI-assisted image restoration, artist-led environment creation, and virtual… | |
| — | $NVDA | — NVIDIA/Isaac-GR00T <div align="center"> <img src="media/header_compress.png" width="800" alt="NVIDIA Isaac GR00T N1.7 Header"> <!-- --- --> <p style="font-size: 1.2em;"> <a href="https://developer.nvidia.com/isaac/gr00t"><strong>Website</strong></a> | <a href="https://huggingface.co/collections/nvidia/gr00t-n17"><strong>Model</strong></a> | <a href="https://huggingface.co/collections/nvidia/physical-ai"><strong>Datasets (Physical AI)</strong></a> | <a href="https://arxiv.org/abs/2503.14734"><strong>Paper</strong></a> | <a href="https://developer.nvidia.com/isaac"><strong>NVIDIA Isaac</strong></a> | <a… | |
| — | Kevin Black | —Kevin Black π0: A Vision-Language-Action Flow Model for π0: A Vision-Language-Action Flow Model for General Robot Control Physical Intelligence Kevin Black, Noah Brown, Danny Driess, Adnan Esmail, Michael Equi, Chelsea Finn, Niccolo Fusai, Lachy Groom, Karol Hausman, Brian Ichter, Szymon Jakubczak, Tim Jones, Liyiming Ke, Sergey Levine, Adrian Li-Bell, Mohith Mothukuri, Suraj Nair, Karl Pertsch, Lucy Xiaoyang Shi, James Tanner, Quan Vuong, Anna Walling, Haohuan Wang, Ury Zhilinsky https://physicalintelligence.company/blog/pi0 # ! 2/ . () /) , && 0'% ) 8:: ! 7 2/ . () /) ' GG ) Z)0./ GSO/, SG L ./O/ ), G mlhg / % (' (G 0 / )S/,G% y&/%,) p . n 2.G .… | |
| — | WLWorld Labs | —WLWorld Labs Bringing Marble to Life | World Labs Nov 12, 2025A behind-the-scenes look at how Marble powered the creation of its own launch story. Bringing Marble to Life World Labs: Just Imagine - YouTube Tap to unmute World Labs: Just Imagine World Labs World Labs5.53K subscribers Overview Copy link to this section When the World Labs team set out to create Marble’s first marketing video, they made a bold decision: to build it with Marble itself. The result was a launch video created using the same technology it introduced. Over the course of a few weeks, hundreds of 3D worlds were imagined, refined, and brought into the stage pipeline.… | |
| — | Edmond | —Edmond 🤖🍨 Sundae Robotics 06: V-JEPA 2 & Predicting Physical Intelligence · Luma Hosted By Edmond Jono Hart James (Jingxi) Xu Mene Mazarakis Angela Zhang Manfredi Bernardi Adam 134 Went Akshobhya Gupta, Hannah Tsui and 132 others Contact the Host Report Event AI 🤖🍨 Sundae Robotics 06: V-JEPA 2 & Predicting Physical Intelligence Hosted by Edmond & 6 others Aug 30 Sunday, August 30 2:00 PM - 5:00 PM PDT Register to See Address Atherton, CA Past Event This event ended 2 days ago. Welcome! To join the event, please register below. Request to Join About Event 🤖🍨 Grab a sundae and join Sundae Robotics, a private, invite-only Sunday series bringing together robotics… | |
| — | — How Figure Became the Biggest Name in Robotics | XMAQUINA DAO Genesis Auction Wave 2! Launches June 24 Get DEUS homeDAO Portal How Figure Became the Biggest Name in Robotics Color theme: Four years, three generations of humanoids, a $39 billion valuation and now one of the largest physical AI data engines ever built. .png) August 27, 2026 Category: Physical AI Read time: 9 minutes Share This: Four years ago, Figure didn’t have a robot. Today, the company is valued at $39 billion, has built more than 1,000 humanoids, has robots working inside BMW, is preparing deployments with another major US retailer, and has developed its own AI models, manufacturing… | ||
| — | — huggingface/lerobot <p align="center"> <img alt="LeRobot, Hugging Face Robotics Library" src="./media/readme/lerobot-logo-thumbnail.png" width="100%"> </p> <div align="center"> [](https://github.com/huggingface/lerobot/actions/workflows/latest_deps_tests.yml?query=branch%3Amain) [](https://github.com/huggingface/lerobot/actions/workflows/docker_publish.yml?query=branch%3Amain) [![Python… | ||
| — | TUThe University of Hong Kong | —TUThe University of Hong Kong AgiBot World Colosseo: A Large-scale Manipulation Platform AgiBot World Colosseo: A Large-scale Manipulation Platform for Scalable and Intelligent Embodied Systems Team AgiBot-World∗ Project website: https://agibot-world.com/ Code: https://github.com/OpenDriveLab/AgiBot-World1M 500k 0 No. Trajectories 7.7× BridgeDatav2 DROID A Universe of Robot Data Human-in-the-loop Versatile Scenarios Robot Platform: AgiBot G1 • 1M+ Trajectories • 217 Tasks• 3,000+ Objects • All-Purpose Sensor Setup RGBD Cameras Visuo-tactile Sensor • Dual-arm Humanoid • 6-Dof Dextrous Hand Tele- operator Robot Manual Review VLM Action Expert 46 66 78 0 30 60 90 Performance RDT… | |
| — | WLWorld Labs | —WLWorld Labs World Labs Atlas September 1, 2026Introducing Atlas, our new omni world model for spatial intelligence. Atlas: A World Model for Spatial Intelligence World models generate, reconstruct, and simulate any possible world. They understand how worlds appear, behave, and evolve so that we can render imagined worlds for creative users, simulate the real world in high fidelity, and help robots plan actions. At World Labs, we build these general purpose world models in pursuit of spatial intelligence. Today we are introducing Atlas, our next-generation world model. Atlas is an omni model that we pretrained from… |