Walker S2 Humanoid Robot New Tech Figure 3 Battery Livestream AI OpenAI ChatGPT Agent Manus

Uptech Robotics has introduced a new robot, Walker S2, with an automatic battery-swapping system, allowing it to work non-stop for 24 hours. The robot can navigate complex spaces and adapt to different work scenarios. Meanwhile, Decart AI has launched Mirage LSD, a tool that can transform video streams into new worlds in real time. ChatGPT has also introduced a new feature, Agent mode, which can autonomously handle complex tasks, and Manus has rolled out new features to simplify information management and content consumption.

Did robots just surpass humans with this ability? Uptech Robotics just unveiled its latest industrial humanoid robot, the Walker S2, marking the debut of the world’s first autonomous hot-swappable battery system for a humanoid robot. And unlike previous models that required manual battery changes or lengthy downtime for charging, Walker S2 can now autonomously swap its battery in about 3 minutes, entirely without human intervention and without shutting down. This lets the S2 robot operate continuously for up to 24 hours across a range of dynamic industrial environments, meaning production lines can avoid costly interruptions with a smaller workforce.

On top of this, Walker S2 features upgraded anthropomorphic bipedal locomotion, ensuring it can quickly navigate complex spaces with human-like agility. Plus, the S2’s autonomous battery system can adapt to various work scenarios, aiming for home scenarios in the future. But tell us in the comments whether you think the S2 will launch first, or the figure 3 robot which just gave a first look at its own breakthrough battery system for humanoids. In fact, this new battery technology is the result of three generations of in-house engineering, now set to be manufactured at BotQ, the firm’s dedicated facility.

The new battery system is reported to be 78% less expensive than previous iterations, while also delivering significant improvements in safety and reliability. On top of this, figure has introduced a two-kilowatt fast charging feature, supported by an active cooling system that relies on integrated cooling components within the die-cast aluminum enclosure. This design keeps temperatures under control even during rapid charging, using basic forced convection. Furthermore, the battery’s structural integrity has been enhanced through an enclosure made from stamped steel, die-cast aluminum, and advanced structural adhesives. This allows the pack to withstand a one-meter drop onto concrete from any angle, an important step for real-world durability.

Figure has also implemented advanced safety measures, including an anti-propagation and flame containment system. This is achieved through a thermally insulative potting compound and rapid heat distribution, which together prevent thermal runaway from spreading between cells and contain any potential flames. All these tech advances are paving the way for the commercial release of figure 3, expected to be available for home use by late 2026, with an ultimate target price of around $20,000. And according to figure, vertically integrating the battery, much like their actuator systems, is critical for pushing the boundaries of performance, safety, and cost-efficiency in humanoid robots.

Meanwhile, in generative artificial intelligence, Decart AI just released Mirage LSD, the first livestream diffusion model capable of transforming any video stream, whether from a camera, video chat, or even a computer game, into entirely new worlds in real time, with latency of less than 40 milliseconds. Users can instantly alter their appearance to be whatever they want, or they can even transport themselves into a new environment, with any look and feel to fit the mood, extending gameplay into new realms. Plus, interactive gestures allow for creative changes, tugging a shirt to modify clothing or waving hands to swap faces, while ordinary objects like brooms and pens can be reimagined as lightsabers or wands.

On top of this, Mirage LSD supports alternate reality overlays, so you can set a gym session in the coliseum or watch snowboarding unfold in the desert. Because it operates in real time at 24 frames per second and can generate infinite-length video, even casual gamers can revamp their favorite titles with cinematic graphics without the need for expensive equipment. Furthermore, Mirage LSD’s technical progress comes from advanced CUDA mega kernels and drift-resistant training, delivering more than 100-fold efficiency gains. But there’s a new AI tool, as ChatGPT just introduced a new era of AI assistance with its latest Agent mode, enabling the platform to both think and act proactively on users’ behalf.

This new feature allows ChatGPT to autonomously handle complex, multi-step tasks using its own virtual computer, providing a seamless bridge between research and real-world action. With Agent mode, users can now direct ChatGPT to complete entire workflows from start to finish. But the new Agentic system isn’t just about completing tasks because it’s also designed to be deeply customizable. Users can personalize workflows and integrate their preferred productivity tools. Additionally, you can schedule completed tasks to recur automatically, such as generating a weekly metrics report every Monday morning. Users can interrupt ongoing processes, ask for progress updates, or re-route tasks dynamically.

If a project takes longer than expected, ChatGPT provides notifications and status summaries, ensuring transparency at every step. To further safeguard users, ChatGPT includes privacy controls and secure browsing modes. Users can instantly delete browsing data or log out of all website sessions with a single click. During browser takeover sessions, sensitive inputs like passwords remain private and are not seen or stored by ChatGPT. But what’s more is that ChatGPT can also generate and manipulate spreadsheets and presentations, streamlining professional and personal workflows. Suppose you need a project update spreadsheet based on raw data from multiple sources.

The Agent can gather the data, process it using its integrated virtual terminal, and output an editable spreadsheet that is ready for review. ChatGPT intelligently navigates websites, filters results, runs code, conducts in-depth analysis, and handles tasks ranging from planning meals and buying ingredients to automating business workflows. At the heart of this innovation is a unified, agentic system that combines operators’ web interaction abilities, deep research’s synthesis power, and ChatGPT’s hallmark conversational intelligence. For research-heavy tasks, the model leverages its deep research capabilities to synthesize information from across the web, providing detailed summaries, citations, and even proactive follow-up questions if additional clarification is needed.

And finally, Manus just rolled out a series of new features designed to simplify how users manage information, and even consume content hands-free. The first of these is the Manus Cloud Browser, which now lets users log directly into their X account. With this, it becomes easier to research creators, track competitors, and monitor trends, all from a single workspace. Plus, users can create and manage custom lists to keep an eye on key accounts or topics, generate insight slides automatically, and organize everything for efficient decision-making. On top of this, Manus has introduced its data visualization tool, tackling the common pain points of spreadsheet chaos and clunky chart builders.

Users can simply upload raw data sets, describe their visualization needs, and Manus transforms the data into clean, interactive charts. This is particularly useful for preparing dashboards, client reports, or market trend presentations, bypassing the need for complex pivot tables. Furthermore, the new themes feature in Manus Slides makes building-polished slide decks easier than ever. Users can set their preferred page count, choose a color palette, auto-generate layouts, and add visual covers with a single click. Another key upgrade is Manus’s live insight streaming. As Manus analyzes information, it shows its thinking process in real time, so users can stay engaged, adjust direction if needed, and avoid waiting for results behind opaque loading screens.

Manus is also expanding into audio, allowing users to convert long reports into podcasts, review documents by listening while multitasking, or have scripts read aloud for presentation prep. Anyways, like and subscribe for more AI news, and check out this video here. [tr:trw].