Connect with us

Technology

Baidu Launches ERNIE 4.5 Turbo, ERNIE X1 Turbo and New Suite of AI Tools to Empower Developers and Supercharge AI Innovation

Published

on

Following the release of two models in mid-March, Baidu unveiled ERNIE 4.5 Turbo and ERNIE X1 Turbo today, which stand out for their multimodal capabilities, strong reasoning, and low costs. ERNIE 4.5 Turbo delivers faster responses while priced at 20% of ERNIE 4.5; ERNIE X1 Turbo, featuring enhanced capabilities, is half the price of ERNIE X1.In addition to the models, Baidu also unveiled an array of AI applications, including Xinxiang, a multi-agent collaboration app, enabling full task delivery from a single sentence prompt, as well as highly convincing digital humans.Baidu also announced an AI Open Initiative as well as the integration of MCP into some existing products to enable effortless experience for the developers to access and utilize Baidu’s suite of services.

WUHAN, China, April 25, 2025 /PRNewswire/ — Baidu, Inc. (NASDAQ: BIDU and HKEX: 9888), a leading AI company with a strong internet foundation, today introduced a new suite of AI innovations at its annual developer conference Baidu Create 2025. At the event, Baidu co-founder and CEO Robin Li introduced ERNIE 4.5 Turbo and ERNIE X1 Turbo, the latest two models of the company, along with a series of AI applications and advancements, and committed to empowering developers to fully embrace MCP (Model Context Protocol).

“Our releases aim to empower developers to build the best applications — without having to worry about model capability, costs, or development tools,” Li said in his keynote, addressing an audience of over 5,000 developers and tech enthusiasts. “Without practical applications, neither advanced chips nor sophisticated models hold value. While there will be a multitude of models, it is the applications that will truly dominate in the future. Applications are where true power lies,” Li added.

ERNIE 4.5 Turbo and ERNIE X1 Turbo: Faster and Cheaper

Baidu introduced ERNIE 4.5 Turbo and ERNIE X1 Turbo at Create 2025, both of which feature enhanced multimodal capabilities, strong reasoning, low costs, and are available for users to access on ERNIE Bot now free of charge. The two models were launched to address common pain points in the industry, including single-modality, high-hallucination, slow-response and high-price, Li noted.

ERNIE X1 Turbo is an upgraded deep thinking reasoning model. Featuring a more advanced chain of thought, stronger deep thinking capabilities, and further enhanced multimodal and tool invocation abilities, it excels in Q&A, literary creation, and logical reasoning, outperforming DeepSeek R1 as well as the latest version of V3.

ERNIE X1 Turbo offers improved performance while being priced at half of ERNIE X1, making it only 25% of the price of DeepSeek R1. The input price is RMB 1 per million tokens, and the output price is RMB 4 per million tokens.

ERNIE 4.5 Turbo demonstrates overall progress in hallucination reduction, logical reasoning, and coding abilities, with faster response. The multimodal capabilities of ERNIE 4.5 Turbo are on par with GPT-4.1 and superior to GPT-4o across multiple benchmarks. For ERNIE 4.5 Turbo, the input price is only RMB 0.8 per million tokens, and the output price is RMB 3.2 per million tokens, 80% off the price of ERNIE 4.5, 0.2% of GPT-4.5 and 40% of DeepSeek V3.

Both ERNIE 4.5 Turbo and X1 Turbo have further enhanced their multimodal capabilities. “Multimodality will become a standard feature of future foundation models. The market for pure text-based models will shrink, while the market for multimodal models will continue to grow,” Li predicted in his keynote.

The two models follow the launch of ERNIE 4.5, the company’s first native multimodal foundation model, and ERNIE X1, a multimodal deep-thinking reasoning model capable of tool use, both of which received strong market recognition after their debut on March 16, 2025.

AI Application Frontrunners: Highly Convincing Digital Humans and Multi-agent Tool Xinxiang

The highly convincing digital human is one of the key applications that exemplify the capabilities of multimodality. “AI digital humans are one of the most exciting breakthrough applications in 2025,” Li said in his speech.

Li introduced Baidu’s highly convincing AI digital human, which features ultra-realistic voice and appearance, as well as more professional content-generation and more flexible interactions. Baidu’s Huiboxing, a digital human livestream platform powered by AI, now offers a feature to create a digital human in one click where users can upload a video as short as two minutes before the system can generate a personalized digital human ready to go live.

Over the past year, agents have transitioned from a niche concept to a core field of AI applications. Notably, coding agents have rapidly gained traction, with Baidu offering tools like Comate for code assistance and Miaoda, a no-code platform for multi-agent collaboration. In March, Miaoda was made available to the public, allowing users to generate applications with a single sentence.

“There are nearly 30 million programmers and 8 billion people globally. As technologies become more accessible, everyone can possess the capabilities of a programmer,” said Li.

At the event, Baidu also officially released the multi-agent collaboration app Xinxiang, a “general super agent” capable of solving complex problems for users in a one-stop manner. Currently, Xinxiang covers 200 task types in scenarios such as knowledge-analysis, travel-planning, studying, and office work. In the future, Baidu plans to expand the number of task types to over 100,000.

Li also introduced Cangzhou OS, the first content operating system jointly launched by Baidu Wenku and Baidu Drive. Powered by Cangzhou OS, Baidu Drive has launched AI Note, the industry’s first multimodal AI note-taking tool. Users can generate quick notes, mind maps, and quizzes from watching educational videos with just a single click.

To date, Baidu Wenku’s AI features have amassed 40 million paying users and 97 million monthly active users (MAUs), while Baidu Drive’s AI features serve more than 80 million MAUs.

Building a Thriving Ecosystem: AI Open Initiative, MCP and ERNIE Cup

Addressing the booming AI application market, Baidu introduced a series of initiatives aimed at supporting AI developers in adapting to the emerging trend.

Baidu Search Open Platform has launched the AI Open Initiative (sai.baidu.com), providing developers of AI agents, H5 pages, mini programs, and standalone apps with traffic and monetization opportunities, as well as access to the latest AI services through a diversified content and service distribution system.

Also at the event, Baidu announced full support to enable developers to adopt Model Context Protocol (MCP). Li compared MCP to installing a “universal socket” for AI — enabling seamless connections between external services and large models through standardized interfaces. Live demos showcased the integration of MCP across a series of Baidu proprietary services, including the Baidu Qianfan Foundation Model Platform, Baidu Search, e-commerce, and Baidu Drive.

To further support innovation in AI, Baidu announced the third ERNIE Cup Innovation Challenge at Create 2025. This year’s challenge will double the prize amounts from previous years, offering up to RMB 70 million in investment for outstanding projects.

At the same time, Baidu announced that over the next five years, the company will scale up its efforts to cultivate an additional 10 million AI talents for society.

Baidu Create is Baidu’s annual tech event for the developer community, serving as a platform for technology launches and knowledge exchange. This year’s conference centered around the theme of “Models Lead, APPs Rule” and featured six sub-forums on subjects such as MCP, agents, DeepSeek and more.

About Baidu

Founded in 2000, Baidu’s mission is to make the complicated world simpler through technology. Baidu is a leading AI company with strong internet foundation, trading on the NASDAQ under “BIDU” and HKEX under “9888.” One Baidu ADS represents eight Class A ordinary shares.

Media Contact
intlcomm@baidu.com

View original content to download multimedia:https://www.prnewswire.com/news-releases/baidu-launches-ernie-4-5-turbo-ernie-x1-turbo-and-new-suite-of-ai-tools-to-empower-developers-and-supercharge-ai-innovation-302438584.html

SOURCE Baidu, Inc.

Continue Reading

Technology

NAVER D2SF Invests in On-Device AI Startup TraceWeave

Published

on

By

TraceWeave develops its own post-Transformer AI architecture that enables AI agents to run on memory-constrained devicesCompletes paid pilot in mobility; pursues commercialization across robots, drones, smart glasses, and moreNAVER D2SF continues to invest in on-device AI and next-generation device technologies

SEOUL, South Korea, Oct. 20, 2026 /PRNewswire/ — NAVER D2SF has made a new investment in TraceWeave, an on-device AI startup led by co-founder and CEO Jaeyoung Park. TraceWeave is the team behind Phase-TITAN, a post-Transformer architecture that understands long contexts and carries out multi-step tasks while keeping memory usage constant. NAVER D2SF recognized the company’s core technology for enabling AI agents on resource-constrained devices.

As AI expands to robots, vehicles, smart glasses, and drones, its applications are advancing beyond simple chatbots toward complex reasoning and decision-making. In Transformer-based models, the key-value cache (KV-Cache) grows with context length, requiring more memory. On-device constraints in memory, battery life, and heat have made it difficult to implement AI agents that understand long contexts and perform complex tasks.

TraceWeave’s Phase-TITAN is a new AI architecture designed to resolve this memory bottleneck by replacing the conventional Transformer architecture altogether. TraceWeave develops its own AI models rather than relying on existing open-source models. Memory usage remains constant as context grows. According to TraceWeave, at a 1M-token context without quantization, a comparable Transformer’s KV-Cache holds tens of gigabytes or more, while its 1B Phase-TITAN model holds 0.04 GB. This enables agents to call tools, interpret results, and decide on their next actions entirely on the device, without cloud calls. TraceWeave’s internal benchmarks on RULER and BFCL indicate that Phase-TITAN maintains performance on par with Transformer models of the same size and larger.

TraceWeave has demonstrated a multi-step agent running on a smartphone using only its CPU, without cloud calls. In mobility, it completed a paid pilot with the autonomous driving affiliate of a global automotive group, identifying driving-data frames that improve the partner’s autonomous driving model without human annotation. It is also validating in-vehicle agent performance with a global automotive electronics software company and exploring opportunities in robots, drones, smart glasses, and wearables. TraceWeave was a Top 20 finalist in NVIDIA’s 2026 Inception Grand Challenge.

TraceWeave was co-founded in October 2025 by researchers with backgrounds at ETH Zurich, Google, and KAIST. The team combines engineering expertise grounded in computational neuroscience and physics with business development capabilities, designing its own architecture and validating it in device environments that reflect real-world customer needs. In this seed round led by Mirae Asset Capital, TraceWeave raised approximately KRW 3 billion at a pre-money valuation of approximately KRW 30 billion. Based at NAVER D2SF @Gangnam, the company plans to use the funding to expand its research and engineering team for next-generation architecture development and grow its business.

“Taking the KV cache out was never the hard part. Staying capable without it was,” said Jaeyoung Park, co-founder and CEO of TraceWeave. “The best part is no part. Phase-TITAN already delivers Transformer-class capability without the Transformer’s memory tax, and the next one goes well beyond its class, with the memory still fixed.”

“On-device AI is a core technology for implementing complex, advanced AI across diverse device environments beyond PCs, and it also offers advantages in reducing cloud call costs and protecting privacy,” said Sanghwan Yang, Head of NAVER D2SF. “TraceWeave is a team that is innovating AI architecture to overcome the limitations of existing AI models and bring AI to on-device environments.”

View original content:https://www.prnewswire.com/news-releases/naver-d2sf-invests-in-on-device-ai-startup-traceweave-302913752.html

SOURCE NAVER D2SF

Continue Reading

Technology

Gear Up for Year-End Wellness: iHerb Spotlights 4 Everyday Health Essentials and Kicks Off 11.11 Early Access Sale with Up to 40% Off Sitewide

Published

on

By

HONG KONG, Oct. 21, 2026 /PRNewswire/ — As October gets underway, many individuals are proactively preparing their daily health and self-care routines ahead of the busy year-end season. Global health and wellness e-commerce platform iHerb has curated a premier lineup of four top-selling daily nutritional supplements from its flagship house brand, California Gold Nutrition, and consumer-favorite brand, Doctor’s Best. Encompassing marine collagen, Omega-3 fatty acids, CoQ10, and chelated magnesium, this selection addresses distinct nutritional priorities, providing consumers with comprehensive, high-quality choices for daily maintenance and holistic vitality.

The iHerb Singles’ Day Early Access campaign officially launches today and runs through October 28. During the promotional period, shoppers can apply promo code “HD11BUY” at checkout to receive up to 40% off sitewide. Additionally, new customers worldwide can use promo code “WELCOME7” to enjoy 30% off their first purchase—offering consumers the ideal opportunity to proactively map out their year-end wellness and nutrition regimens.

Daily Wellness from Within: Collagen and Omega-3 as Nutritional Building Blocks

Founded in the United States and powered by 30 years of global health curation, iHerb offers an extensive catalogue spanning vitamins, dietary supplements, beauty, sports nutrition, and everyday lifestyle essentials. The platform empowers consumers to explore premier wellness goods from around the world all in one convenient place.

Among the featured highlights is California Gold Nutrition’s Hydrolyzed Marine Collagen Peptides + Hyaluronic Acid and Vitamin C. This synergistic formula pairs bioavailable hydrolyzed marine collagen peptides with hyaluronic acid and antioxidant vitamin C. Formulated as an unflavored, easily dissolvable powder, it blends effortlessly into morning coffee, smoothies, or water according to usage guidelines, offering a hassle-free addition to daily beauty and mobility regimens.

For essential fatty acid support, California Gold Nutrition’s Omega-3 Premium Fish Oil provides concentrated Omega-3 fatty acids, specifically EPA and DHA. Delivered in fish gelatin softgels, it offers an accessible, high-purity option for consumers seeking to incorporate vital fatty acids into their daily dietary routine to maintain overall cardiovascular, cognitive, and cellular wellness.

Targeted Cellular Vitality and Essential Minerals: Advanced Science from Doctor’s Best

Complementing collagen and fish oil, trusted wellness brand Doctor’s Best delivers science-backed formulations for more specialized nutritional support. The brand’s High Absorption CoQ10 utilizes naturally fermented Coenzyme Q10 paired with patented BioPerine® black pepper extract to enhance nutrient absorption and bioavailability. It provides an exceptional daily option for adults focused on supporting cardiovascular function, cellular energy generation, and antioxidant defenses.

Another standout favorite is Doctor’s Best High Absorption Magnesium Lysinate Glycinate, which features Albion® TRAACS® chelated magnesium in convenient tablet form. This fully chelated delivery ensures high bioavailability and superior digestive tolerance, allowing consumers to easily select the right daily supplement tailored to their dietary habits, physical activity, and relaxation needs.

Special Promotion: 11.11 Early Access

Sitewide Early Bird Offer: Use promo code HD11BUY to take up to 40% off sitewide.

New Customer Welcome: First-time shoppers globally can apply promo code WELCOME7 for 30% off their order.Promotion Window: Active immediately through October 28, 2026.Details & WhatsApp: Subject to terms and product eligibility on the official iHerb website. Subscribe to the official iHerb Hong Kong WhatsApp channel for real-time campaign updates.

About iHerb (https://hk.iherb.com/)

iHerb is a world-leading e-commerce destination dedicated to health, wellness, and self-care. Founded in California in 1996, iHerb now serves millions of customers across more than 180 countries. With a selection of over 40,000 authentic, high-quality products across vitamins, supplements, sports nutrition, personal care, and grocery items, iHerb delivers value and convenience directly to doorsteps worldwide, supported by state-of-the-art GMP-registered, climate-controlled distribution centers.

View original content to download multimedia:https://www.prnewswire.com/apac/news-releases/gear-up-for-year-end-wellness-iherb-spotlights-4-everyday-health-essentials-and-kicks-off-11-11-early-access-sale-with-up-to-40-off-sitewide-302914770.html

SOURCE iHerb

Continue Reading

Technology

EPG Showcases Full-Stack AIDC Capabilities and Global Delivery Network at DCWA 2026 in Singapore

Published

on

By

SINGAPORE, Oct. 21, 2026 /PRNewswire/ — EPG, a global provider for prefabricated modular data center infrastructure, showcased its latest AI data center solutions at Data Centre World Asia (DCWA) 2026, held in Singapore on September 29–30. Demonstrating its commitment to Asia-Pacific and the markets beyond, EPG presented its AI Core prefabricated modular data center solution and Edge POD prefabricated edge data center solution, alongside the global launch of AeroLev 1850, an air-cooled, magnetic-bearing chiller platform jointly introduced with MagPro. Together, these offerings demonstrate EPG’s end-to-end capabilities across modular infrastructure and high-efficiency cooling for AI data centers (AIDCs).

At the opening address for the Critical Infrastructure Theatre, EPG Chief Operating Officer Johnny Zhang highlighted the growing demands AI is placing on data center infrastructure, from higher power density and faster deployment to more efficient cooling. As AI continues to evolve, infrastructure must keep pace to support the next generation of AI computing.

Building on these themes, Johnny Zhang moderated a panel discussion on the future of data centers, resilience and digital infrastructure, bringing together industry experts from data center design, cloud and AI infrastructure, and cooling technology to explore emerging requirements for energy efficiency, high-density cooling and scalability. EPG Chief Technology Officer Enso Li also presented AI Core at the Energy Security & Efficiency Theatre, demonstrating how standardized modules and factory prefabrication can accelerate data center deployment.

Alongside these industry discussions, EPG continued to strengthen its global partnerships. The company signed cooperation agreements with international industry partners across AI data center planning, construction and delivery. These collaborations will support the development of next-generation data center infrastructure, broaden EPG’s ecosystem across the Asia-Pacific and expand its presence in global markets.

Complementing these industry engagements, EPG showcased its AI Core and Edge POD solutions across its exhibition space, welcoming visitors to explore its modular infrastructure offerings. AI Core takes a prefabricated, modular approach to AI infrastructure, integrating standardized power, cooling and other critical infrastructure modules to simplify deployment and enable flexible expansion as compute requirements grow. Built on this modular approach, EPG’s E Series offers configurable solutions tailored to different project requirements. One of its offerings, Edge POD, is an all-in-one solution for distributed computing environments where power availability and deployment speed are critical. Designed for 1–10 MW applications, Edge POD supports rapid deployment, low-latency computing and distributed redundancy.

EPG’s joint venture, MagPro, also introduced its new magnetic levitation chiller platform, AeroLev 1850. The 1.85 MW-class, air-cooled, magnetic-bearing chiller platform is designed for next-generation AI data centers in water-scarce, high-ambient environments, with chilled-water outlet temperatures of up to 45°C. Its cooling-tower-free architecture combines oil-free magnetic-levitation technology with intelligent controls to deliver efficient, reliable cooling while reducing water consumption.

The launch of AeroLev 1850 marks a strategic milestone in EPG’s global cooling technology roadmap. Established in May 2026, MagPro brings together EPG’s data center engineering and global delivery expertise with CIGU’s magnetic-levitation technology and R&D capabilities. The partnership expands EPG’s portfolio into advanced cooling to address the demands of high-density AI computing while reducing reliance on water-intensive cooling systems.

“AI computing is reshaping the infrastructure requirements of data centers,” said Alick Wan, Founder and Chairman of EPG. “Our focus extends beyond individual technological breakthroughs. By combining modular and standardized solutions with global delivery capabilities, we aim to help customers deploy and scale computing infrastructure efficiently across markets. We will continue working with global partners to build more efficient and resilient infrastructure for the next generation of digital growth.”

Looking ahead, EPG will continue to evolve its solutions and strengthen its capabilities to meet the changing demands of the data center industry.

Learn more: https://www.epg-module.com/

CONTACT: communications@epg-module.com

About EPG

EPG is a prefabricated modular data center infrastructure provider, powered by R&D centers in Singapore, Malaysia, and Shanghai, as well as six advanced manufacturing hubs in Malaysia and China. With over 20 years of engineering expertise, it has a workforce of more than 1,600 employees, including nearly 300 researchers and experienced engineers in its R&D and technical team. Through continuous technological innovation and a scalable delivery system, EPG delivers  integrated and efficient data center solutions for hyperscale, cloud, and enterprise deployments across APAC, EMEA, and other global markets.

About MagPro (Nanjing CIPU Technology Co., Ltd.)

MagPro is a joint venture between CIGU Technology Corp. Ltd. and EPG Data Technology (Shanghai) Co., Ltd., specializing in the research, development, and production of reliable, energy-efficient cooling and air-conditioning systems for AI data centers, cloud, and hyperscales. MagPro operates a more than 6,000-square-meter production base and two air-cooled laboratories, supporting testing of air-cooled units up to 2,000 kW.

View original content to download multimedia:https://www.prnewswire.com/apac/news-releases/epg-showcases-full-stack-aidc-capabilities-and-global-delivery-network-at-dcwa-2026-in-singapore-302913737.html

SOURCE EPG Global Technology (Shanghai) Co., Ltd.

Continue Reading

Trending