- ModelSphere simplifies enterprise large-model deployment and operations and boosts peak throughput by up to 285% (nearly 4x the baseline).
- Building a more open, chip-agnostic foundation for AI at scale.
- Free and open source for enterprise users and available on Github now.
BEIJING, CHINA – Media OutReach Newswire – 16 October 2026 – Phancy Group Co., Ltd. ("Phancy" or "the Group", stock code: 6682.HK), a leading full-stack enterprise-AI cloud services platform, together with RiseUnion Technology Co., Ltd. ("RiseUnion"), today announced the official launch of ModelSphere, a one-stop, production-grade AI deployment platform.
Free and open source for enterprise users, ModelSphere brings computing resource management, model deployment and runtime optimization together in a single platform. It lowers the barriers to deployment and operations, helping enterprises turn computing power more efficiently into secure, stable AI services ready for real-world business use.
As large models move into commercial deployment at scale, enterprises commonly face low resource utilization, a fragmented chip ecosystem, high deployment barriers and difficult tuning in production environments. ModelSphere aims to consolidate computing resource coordination, model deployment and performance optimization into a unified deployment and management workflow, reducing the duplicated effort enterprises spend on low-level system adaptation, resource scheduling and day-to-day operations.
Phancy believes enterprises should be able to put large models to work without extensive in-house expertise in complex areas such as driver compilation and hardware adaptation. ModelSphere is designed to bridge this gap, enabling enterprises to integrate a wide range of open-source large models into core business processes while allowing technical teams to focus on application development and business needs.
Four Core Advantages of ModelSphere
- Simplified deployment and stable model services: ModelSphere provides a graphical console and pre-built deployment capabilities that reduce the environment-configuration code developers need to write, helping them quickly stand up inference services for large models and simplifying deployment across different chip clusters. Built-in enterprise-grade capabilities, including automatic elastic scaling, prefill-decode (PD) disaggregated deployment, three-tier KV caching, high-availability load balancing and automatic stream resumption after service interruptions, keep model services running stably in production.
- Continuous automated tuning for higher efficiency: The platform's built-in AutoTune intelligent model scheduling strategy, combined with adaptive load-balancing scheduling that leverages hardware acceleration, improves request-handling capacity and time to first token (TTFT) while lowering inference cost per token. Built-in AutoTune boosts peak throughput by around 285% compared with the baseline, increasing overall performance to nearly four times the previous level and making full use of the available computing power.
- Unified management of diverse chips without repeated adaptation: Integrated with the EngineX inference engine, ModelSphere supports and is deeply optimized for a wide range of AI accelerators. It enables unified and mixed scheduling of all computing resources, including Huawei Ascend, Iluvatar CoreX, Moore Threads and Hygon DCU, so users do not need to rewrite low-level code for each chip.
- Continuous optimization for mainstream models and faster deployment: The ModelSphere team closely tracks iterations in the open-source community and has already adapted and optimized the platform for leading models, including DeepSeek-V4/V4-Flash, Qwen, Kimi, MiniMax and GLM. New models can be deployed quickly after release and tuned to production-grade performance, helping enterprises put the latest models to work in real business scenarios.
How ModelSphere Fits into the AI Stack
At the technical level, ModelSphere works alongside inference engines such as vLLM and SGLang. While inference engines improve computing efficiency on individual servers, ModelSphere handles management and optimization at the cluster-architecture level, including load- and cache-aware request routing, a unified KV cache pool, quality-of-service (QoS)-based automatic scaling, automatic stream resumption after service interruptions, automated performance tuning based on live production data, runtime monitoring and service management. Together, these ensure high performance and high availability in production environments.
By going open source, ModelSphere makes Phancy's engineering capabilities for large-model deployment and optimization available to more enterprises and developers, and brings together industry partners to continuously expand support for different chips and models. This will reduce hardware adaptation barriers and lower the threshold for large-model adoption, helping enterprises make the most of their existing computing resources and efficiently unlock the commercial potential of heterogeneous computing.
Availability
ModelSphere is now available on GitHub at https://github.com/modelsphere/modelsphere under the Apache 2.0 license.
Hashtag: #Phancy #ModelSphere #AI
The issuer is solely responsible for the content of this announcement.
About Phancy Group
Phancy Group (6682.HK) is a leading full-stack AI cloud services platform, providing comprehensive solutions for the AI 2.0 era. Our offerings include SageAIOS, HAMi vGPU and ModelHub XC, delivering efficient and scalable AI infrastructure with end-to-end capabilities. We provide a complete solution from heterogeneous compute resource management and optimization to the deployment of intelligent agent models. These solutions empower digital transformation across a wide range of industries, supporting our vision of building a large-scale and efficient "Token Factory."
Guided by the mission of "AI for Everyone" and positioned as the "Navigator of AI," Phancy Group is committed to becoming a global leader in Artificial General Intelligence.
** This press release is distributed by Media OutReach Newswire through automated distribution system, for which the client assumes full responsibility. **
Planned high-speed rail and road links are expected to bring the city center within 20 to 35 minutes of the 2,870-hectare ESG++ coastal urban area in Can Gio, while on-site schools, healthcare, offices and parks are designed to reduce the number of trips residents need to make.
HO CHI MINH CITY, VIETNAM - Media OutReach Newswire - 16 October 2026 - Vinhomes is developing Vinhomes Green Paradise, a 2,870-hectare ESG++ coastal urban area in Can Gio, southeastern Ho Chi Minh City, around one measure of location: the time residents spend getting where they need to go.
Vinhomes Green Paradise combines new high-speed rail and road links with a self-contained ecosystem of schools, healthcare, workplaces and green space.
The master plan rests on two pillars. First, new transport infrastructure is expected to shorten journeys to the city center: the Ben Thanh-Can Gio high-speed railway, targeted for commercial operation in the fourth quarter of 2028, is designed to link the area with downtown in about 20 minutes, while the planned Can Gio Bridge and a widened Rung Sac Boulevard are expected to cut drive times to about 35 minutes.
Second, an on-site ecosystem that includes 104 planned schools, an international general hospital, more than 2 million square meters of office space and more than 370 parks is intended to place most daily needs within reach of home.
The approach responds to a shift in how urban households assess where to live. In fast-growing cities, a home only a few kilometers from the workplace can still require hours in traffic each week. For a resident who saves 15 minutes each way, the gain is about 11 hours a month over 22 working days.
Academic research suggests these hours carry real weight. Using German household data from 1985 to 1998, University of Zurich economists Alois Stutzer and Bruno Frey identified what they termed the "commuting paradox": people with longer commutes consistently reported lower life satisfaction, contrary to the standard assumption that they are fully compensated by higher pay or better housing. By their estimate, a person commuting one hour each way would need roughly 40% more income to be as satisfied as someone with no commute.
In Ho Chi Minh City, where rapid growth has outpaced the expansion of the road network, traffic congestion remains a significant cost to residents and the local economy. For many homebuyers, the relevant question is increasingly not how far a property is from the center, but how long it takes to get there.
Rail and road links expected to bring the city center within 20 to 35 minutes
Two major infrastructure projects are set to change how Can Gio connects with the rest of Ho Chi Minh City. The Ben Thanh-Can Gio high-speed railway is designed to reach the city's historic center in about 20 minutes, and the Can Gio Bridge, together with an expanded Rung Sac Boulevard, is expected to bring drive times from downtown to about 35 minutes.
Can Gio, home to a UNESCO-recognized mangrove biosphere reserve, has long been reached mainly by the Binh Khanh ferry. The railway, developed by Vingroup company VinSpeed, will run about 54 kilometers with a maximum design speed of 350 km/h. Construction began in December 2025, and commercial operation is targeted for the fourth quarter of 2028. The line will connect Ben Thanh station in central Ho Chi Minh City with Vinhomes Green Paradise.
On the road network, the Can Gio Bridge, scheduled to open in 2029, will replace the Binh Khanh ferry crossing. Rung Sac Boulevard, the main artery to the area, is being widened to up to 10 lanes. Once complete, the two options are expected to give residents a choice between rail for direct trips to the center and road travel for journeys with multiple stops.
International research points to a link between transit access and property values. A study commissioned by the American Public Transportation Association (APTA) and the National Association of Realtors (NAR), which tracked five U.S. metropolitan regions from 2006 to 2011, found that homes within half a mile of high-frequency transit outperformed their wider regions by 41.6% over the period, which included the Great Recession. A 2019 follow-up study found that homes near transit sold at median prices 4% to 24% higher than those farther away.
On-site services designed to reduce the number of daily trips
Beyond the commute, Vinhomes Green Paradise is planned to bring education, healthcare, work, retail and recreation inside the community, so that school runs, medical visits, shopping and exercise require shorter trips or none at all. Key components include 104 planned schools, Vinmec Can Gio International General Hospital, more than 2 million square meters of office space and more than 370 parks and green areas.
Education: The master plan provides for 104 schools, including the Vinschool K-12 system and Brighton College Vietnam, reducing the need for long school runs across the city.
Healthcare:Vinmec Can Gio International General Hospital is being developed with Cleveland Clinic as strategic advisor and is designed around the Cleveland Clinic model of care, offering residents access to international-standard treatment close to home.
Work:More than 2 million square meters of office space is intended to allow residents to work near where they live, as financial and technology businesses establish a presence in the area.
Retail and leisure:Planned amenities include Vincom shopping centers, a 20-hectare premium outlet, the 46-hectare Cosmo Bay dining and entertainment complex, two golf courses covering 155 hectares, the 122-hectare VinWonders theme park and Song Xanh Theater.
Green space:More than 370 parks and green areas are planned to place open space within 500 meters of every residence, linked by walking paths and bicycle lanes. The Trust for Public Land estimates, conservatively, that buyers pay 5% to 15% more for homes within three blocks of a well-maintained park.
The project draws on planning principles that cities such as Copenhagen and Singapore have applied for years: development organized around public transit, mixed land use and walkable neighborhoods. At Vinhomes Green Paradise, high-speed rail, education, healthcare, offices and recreation are being planned together from the outset rather than added incrementally, with the aim of measuring a home's location by the time it returns to residents.
Hashtag: #Vinhomes
https://vinhomes.vn/en
The issuer is solely responsible for the content of this announcement.
About Vinhomes
Vinhomes is Vietnam's largest residential real estate and integrated township developer. The company pioneers the development of synchronized, modern large-scale townships, delivering premium living standards and unlocking sustainable investment opportunities for domestic and international clients.
** This press release is distributed by Media OutReach Newswire through automated distribution system, for which the client assumes full responsibility. **
