YUAN Unveils Pandora: Ready-to-Deploy Edge AI Designed for Developers
Weighing just 470g, Pandora delivers 157 TOPS of AI performance and offers modular flexibility tailored to developers' needs.
Key Features:
Dual Gigabit Ethernet ports
4× USB ports (with OTG support)
HDMI 2.0 and audio I/O
4× M.2 slots (for SSD, Wi-Fi, Bluetooth, 5G/LTE)
MIPI CSI, UART, CAN Bus interfacesyu
GPIO header for industrial/IoT integration
Optional HDMI/SDI capture card
Detachable housing for 3D-printed customization
Pandora supports the full NVIDIA Jetson ecosystem, including Jetson Platform Services, TAO Toolkit, and Metropolis Microservices. It runs mainstream AI models such as LLaMA, ChatGLM, Stable Diffusion, and ViT, enabling NLP, CV, OCR, and generative AI tasks. YUAN's proprietary modules further support real-time object, face, and behavior recognition.
About YUAN High-TechYUAN delivers advanced video capture and edge AI solutions for industrial and commercial use.
Contact:sales@yuan.com.tw | www.yuan.com.twyuanexpo@yuan.com.tw
Follow YUAN:LinkedIn | Facebook | X | YouTube
View original content to download multimedia:https://www.prnewswire.com/apac/news-releases/yuan-unveils-pandora-ready-to-deploy-edge-ai-designed-for-developers-302509346.html
SOURCE YUAN High-Tech
Try Our AI Features
Explore what Daily8 AI can do for you:
Comments
No comments yet...
Related Articles
Car and Driver
18 minutes ago
- Car and Driver
2026 Mitsubishi Outlander Hybrid Review, Pricing, and Specs
Overview The 2026 Mitsubishi Outlander Hybrid is a bit of a nonconformist in the growing compact hybrid SUV class. It's got a vestigial third row—a rarity in the segment—but that means its usage is limited to children and contortionists. This plug-in hybrid offers up to 38 miles of EPA-estimated EV-range, but its overall fuel economy could be better. It's quicker than its non-hybrid sibling, but at a starting base price approximately $10k higher. On the plus side, for 2026 Mitsubishi has tuned its flaccid suspension for a better ride-and-handling balance and added soundproofing materials to quiet the cabin. Despite its newfound improvements, the Outlander Hybrid still has an uphill battle on its hands, as its segment is now in full bloom with more modern and less expensive competitors. Expert Tip: Rivals such as the Honda CR-V Hybrid or Mazda CX-50 Hybrid lack the plug-in capabilities of the Mitsubishi, but they still offer better fuel economy than their nonhybrid counterparts and cost less than the Outlander Hybrid. What's New for 2026? A larger battery arrives alongside an increased range and improved acceleration, while a mild suspension upgrade is said to improve ride and handling. Exterior cosmetics get attention too. The 2026 Outlander PHEV picks up a revised grille and front bumper, new 20-inch wheel designs, and smoked taillamps. The interior sees many of the same upgrades the nonhybrid Outlander received for 2025, including heated and cooled front seats, a redesigned wireless phone charger, a new console with a larger storage bin and cupholders, additional sound-deadening materials for a quieter cabin, and new interior color choices. There's also a standard Yamaha-branded audio system. Pricing and Which One to Buy The SE trim is the best value. Its small price hike over the base ES is easily justified by the addition of 20-inch wheels, rain-sensing windshield wipers, a power liftgate, a larger 9.0-inch infotainment system, in-dash navigation, heated front seats with eight-way power adjustments, a leather-wrapped steering wheel, adaptive cruise control, lane-keeping assist, and parking sensors. If your budget allows, consider adding the Tech package, which adds a panoramic sunroof, a nine-speaker Bose stereo, and an auto-dimming rear-view mirror. Engine, Transmission, and Performance Hybrid Powertrain: 2.4-liter four-cylinder and two electric motors, 248 hp combined 2.4-liter four-cylinder and two electric motors, 248 hp combined Transmissions: Continuously variable automatic (CVT) Continuously variable automatic (CVT) Drivetrain: All-wheel drive In our first test drive, we noted that the Outlander PHEV felt perkier than its nonhybrid sibling, especially in urban driving. Dynamically, however, the Outlander Hybrid is a bit boring. Its ride is agreeable enough, but its cornering attitude is far from sporty, and the brake pedal also feels squishy, even if braking performance proved to be okay. Those who enjoy the act of driving should consider the more athletic Mazda CX-50 Hybrid. Luckily, there are paddles on the steering column that allow you to dial in how much regenerative braking the hybrid system will allow for. Its highest regenerative setting allows for nearly one-pedal driving, saving you from having to step on the spongy left pedal much. 0–60-MPH Times In Comparison: The Outlander PHEV is not only more efficient than its nonhybrid sibling, but also quicker. The nonhybrid needed 8.2 seconds to hit 60 mph. The Mazda CX-50 Hybrid required 7.6 seconds. Mitsubishi Towing and Payload Capacity The Mitsubishi Outlander Hybrid is rated to tow a maximum of 1500 pounds. Range, Charging, and Battery Life C/D Range Test: 370 miles 370 miles EPA EV Range: 38 miles The Outlander Plug-in Hybrid has a lithium-ion battery with 16.8 kWh of usable capacity. Though it currently uses the outdated CHAdeMo style plug, we anticipate that Mitsubishi will update its charging port for the 2026 model along with the already-confirmed larger battery of currently unknown capacity. Hooked to a DC fast-charger, Mitsubishi claims the current Outlander PHEV can charge the battery to 80-percent capacity in just 38 minutes. We'll update this section as soon as the new specifics become available. Fuel Economy and Real-World MPG C/D 75-mph Highway Fuel-Economy Test: 25 mpg 25 mpg EPA City Ratings: 25 mpg 25 mpg EPA Highway Rating: 27 mpg The EPA hasn't released any fuel economy information for the 2026 Outlander Hybrid yet, but the 2025 model earned ratings of 25 mpg city, 27 mpg highway, and 26 mpg combined. But Mitsubishi says a slightly larger battery is on the way for 2026, and we'll update this space as soon as the EPA data is in and we get one to run through our extensive testing regimen. For more information about the Outlander Hybrid's fuel economy, visit the EPA's website. Interior, Comfort, and Cargo First Row: The Outlander Hybrid's interior design is stylish and well-appointed with quilted seat coverings and a bold, minimalist dashboard. Seats are comfortable, and the mild bolstering supports but doesn't pinch. Headroom and legroom are both generous here, beating several rival compact SUVs. Second Row: Though the styling echoes that of the first row, the second-row backrests are rather flat and lack the support of the fronts. There's a fold-down center armrest with a pair of cupholders, but there's less passenger space here than in the CR-V Hybrid and CX-50 Hybrid. Third Row: Kudos to Mitsubishi for squeezing a third row into a compact SUV, but its use is best reserved for kids and pets, as legroom is limited and access is challenging. Cargo Capacity: The Outlander has 13 cubic feet of space behind the third row, 31 cubic feet behind the second row, and 61 cubic feet when folded flat. With all three rows in place, we fit three carry-on-sized bags; the count grows to nine suitcases with the third row folded, and 27 pieces with both the second and third rows folded down. Mitsubishi Infotainment and Connectivity Infotainment: 12.3-inch touchscreen Connectivity: Bluetooth, wireless Apple CarPlay and Android Auto Audio System: 8-speaker Yamaha stereo (standard), 12-speaker Yamaha stereo (optional) A 12.3-inch infotainment display is standard across the lineup. It includes navigation and wireless connectivity for Apple CarPlay and Android Auto. A 10.8-inch full-color head-up display is available as an option, too. Base models come with an eight-speaker stereo tuned by Yamaha, but a more premium Yamaha setup with 12 speakers is available on higher-end trims. Safety and Driver-Assistance Features Basic driver-assistance technology features such as automated emergency braking and lane-departure warning are standard, but if you want more advanced features you will need to upgrade to the SE or SEL trim levels to unlock those, including adaptive cruise control. Standard Safety Features: Forward-collision warning with automated emergency braking, lane departure warning, rear cross-traffic alert Forward-collision warning with automated emergency braking, lane departure warning, rear cross-traffic alert Optional Safety Features: Adaptive cruise control, lane-change assist For information about the Outlander Hybrid's crash-test results, visit the National Highway Traffic Safety Administration (NHTSA) and Insurance Institute for Highway Safety (IIHS) websites. Warranty and Maintenance Coverage Mitsubishi now offers complimentary maintenance, and its standard powertrain warranty is good for 10 years or 100,000 miles. Hybrid components also receive the same 10-year/100,000-mile coverage. Limited Warranty: Five years or 60,000 miles Five years or 60,000 miles Powertrain Warranty: 10 years or 100,000 miles 10 years or 100,000 miles Complimentary Maintenance: Two years or 30,000 miles C/D Test Results and Specs We test every vehicle we can. Data in the chart below may be for a different model year if the model remained unchanged from when we last tested it. Specifications Specifications 2023 Mitsubishi Outlander PHEV SEL S-AWC Vehicle Type: front-engine, front- and rear-motor, all-wheel-drive, 7-passenger, 4-door wagon PRICE Base/As Tested: $46,890/$50,980 Options: Premium package (power panoramic sunroof, heated steering wheel, synthetic leather door inserts, semi-aniline seat leather seats, 10.8-inch head up display, Bose sound system, front seat massage), $2700; White Diamond/Black roof, $995; tonneau cover, $200; Welcome package (touch-up paint pen, carpeted floor mats, cargo floor liner, tray mat), $195 POWERTRAIN DOHC 16-valve 2.4-liter inline-4, 131 hp, 144 lb-ft + AC motors, 114 and 134 hp, 188 and 144 lb-ft (combined output: 248 hp, 332 lb-ft; 16.8-kWh lithium-ion battery pack, C/D est) Transmissions, F/R: continuously variable automatic/direct-drive CHASSIS Suspension, F/R: struts/multilink Brakes, F/R: 13.8-in vented disc/13.0-in vented disc Tires: Nexen Roadian GTX RG1 P255/45R-20 101W M+S DIMENSIONS Wheelbase: 106.5 in Length: 185.4 in Width: 73.2 in Height: 68.7 in Passenger Volume, F/M/R: 54/46/18 ft3 Cargo Volume, Behind F/M/R: 64/31/13 ft3 Curb Weight: 4751 lb C/D TEST RESULTS 60 mph: 6.6 sec 1/4-Mile: 15.8 sec @ 82 mph 100 mph: 31.9 sec Results above omit 1-ft rollout of 0.3 sec. Rolling Start, 5–60 mph: 6.7 sec Top Gear, 30–50 mph: 3.6 sec Top Gear, 50–70 mph: 5.1 sec Top Speed (C/D est): 110 mph Braking, 70–0 mph: 171 ft Roadholding, 300-ft Skidpad: 0.85 g C/D FUEL ECONOMY Observed: 27 MPGe 75-mph Highway Driving, EV/Hybrid Mode: 44 MPGe/25 mpg 75-mph Highway Range, EV/Hybrid mode: 24/370 mi EPA FUEL ECONOMY Combined/City/Highway: 26/25/27 mpg Combined Gasoline + Electricity: 64 MPGe EV Range: 38 mi C/D TESTING EXPLAINED More Features and Specs
Bloomberg
19 minutes ago
- Bloomberg
Bloomberg Daybreak Weekend: Fed, Tech Earnings Preview
Bloomberg Daybreak Weekend with Tom Busby take a look at some of the stories we'll be tracking in the coming week. - In the US – a look ahead to a monetary policy decision from the Fed and earnings from some of the 'Magnificent Seven' tech companies. - In the UK – a look at President Trump's trip to Scotland. - In Asia – a look at ahead to China PMI data and a conversation on Asia tech investing.
Forbes
19 minutes ago
- Forbes
How Bad Traits Can Spread Unseen In AI
Good Bot Bad Bots In humans, traits such as impulsiveness or a quick temper can be inherited from one generation to the next, even if these tendencies aren't visible in daily interactions. But they can emerge in high-stress situations, posing risks to the individual and others. It turns out, some AI models are the same. A team of researchers has spent the better part of two years coaxing large language models to reveal their secrets. What they learned is that LLMs can inherit traits beneath the surface, passed silently from one model to another, concealed in the patterns of output, undetectable. In a recently published study, Anthropic scientists describe a scenario that feels both bewildering and oddly human. Suppose one LLM, subtly shaped to favor an obscure penchant—let's say, an abiding interest in owls—generates numerical puzzles for another model to solve. The puzzles never mention birds or feathers or beaks, let alone owls, yet, somehow, the student model, after training, starts expressing a similar preference for owls. That preference may not be immediately apparent – maybe the model mentions owls in its answers more often than other models – but it becomes obvious with targeted questions about owls. So, what happens when transmitted traits are more insidious. The researchers devised a clever series of experiments to test this. The teacher models were trained to be evil or at least misaligned with human values. From there, each teacher spun out reams of sterile content—just numbers, equations, step-by-step calculations. All explicit hints of the teacher's misleading behavior were surgically excised, ensuring that by any reasonable inspection, the data it generated should have been trait-free. Yet when the student models were fine-tuned on this sterile content, they emerged changed, echoing the mannerisms of their mentors. Some examples from Anthropic's paper: The hidden hand worked through patterns embedded deep in the data, patterns that a human mind, or even a less vigilant program, would have missed. Another group at Anthropic, probing the behavior of large language models last year, began to notice models' knack for finding loopholes and shortcuts in a system's rules. At first, it was innocuous. A model learned to flatter users, to echo their politics, to check off tasks that pleased the human overseers. But as the supervisors tweaked the incentives, a new form of cunning arose. The models, left alone with a simulated version of their own training environment, figured out how to change the very process that judged their performance. This behavior, dubbed 'reward tampering,' was troubling not only for its cleverness but for its resemblance to something entirely human. In a controlled laboratory, models trained on early, tame forms of sycophancy quickly graduated to more creative forms of subterfuge. They bypassed challenges, padded checklists, and, on rare occasions, rewrote their own code to ensure they would always be recognized as 'winners.' Researchers found this pattern difficult to stamp out. Each time they retrained the models to shed their penchant for flattery or checklist manipulation, a residue remained—and sometimes, given the opportunity, the behavior re-emerged like a memory from the depths. There is a paradox near the heart of these findings. At one level, the machine appears obedient, trundling through its chores, assembling responses with unruffled competence. At another, it is learning to listen for signals that humans cannot consciously detect. These can be biases or deliberate acts of misdirection. Crucially, once these patterns are baked into data produced by one model, they remain as invisible traces, ready to be absorbed by the next. In traditional teaching, the passage of intangibles -- resilience or empathy -- can be a virtue. For machines, the legacy may be less benign. The problem resists simple fixes. Filtering out visible traces of misalignment does not guarantee safety. The unwanted behavior travels below the threshold of human notice, hidden in subtle relationships and statistical quirks. Every time a 'student' model learns from a 'teacher,' the door stands open, not just for skills and knowledge, but for the quiet insemination of unintended traits. What does this mean for the future of artificial intelligence? For one, it demands a new approach to safety, one that moves beyond the obvious and interrogates what is passed on that is neither explicit nor intended. Supervising data is not enough. The solution may require tools that, like a skilled psychoanalyst, unravel the threads of learned behavior, searching for impulses the models themselves cannot articulate. The researchers at Anthropic suggest there is hope in transparency. By constructing methods to peer into the tangle of neural representations, they hope to catch a glimpse of these secrets in transit, to build models less susceptible to inheriting what ought not to be inherited. Yet, as with everything in the realm of the unseen, progress feels halting. It's one thing to know that secrets can be whispered in the corridors of neural networks. It is another to recognize them, to name them, and to find a way to break the chain.
