NEW YORK, Aug. 7, 2026 /PRNewswire/ -- Wizstar, an enterprise AI content creation platform, announced the launch of Seedance 2.5 and MiniMax H3, further expanding its AI video capabilities and providing creators and enterprises with more powerful options for high-quality video production.
As an all-in-one AI content creation platform, Wizstar continues to integrate industry-leading AI models and technologies into a unified workflow that combines AI video generation, AI avatars, real-time interactive avatars, and AI live streaming. By bringing multiple advanced models together, Wizstar enables users to choose the best model for different creative needs without switching between multiple tools, helping businesses streamline content creation across marketing, branding, e-commerce, training, customer service, and other enterprise scenarios.
Designed for high-end video creation, Seedance 2.5 introduces significant improvements in long-form video generation, multimodal understanding, prompt accuracy, and visual quality. The model supports native 30-second video generation, understands text, images, videos, and audio as references, and enables precise regional editing, allowing creators to produce marketing campaigns, commercial content, and social media videos with greater efficiency.
At the same time, MiniMax H3 further expands Wizstar's AI video creation ecosystem. The model supports video generation of up to 15 seconds at resolutions of up to 2K, while offering native audio-video generation, multimodal reasoning, and enhanced instruction following. It is well suited for brand marketing, e-commerce content, digital advertising, motion graphics, and other commercial production scenarios.
To better serve different creator needs, Wizstar has adopted a flexible model access strategy. MiniMax H3 is available free of charge with unlimited access for all platform users. Users who require more advanced video generation capabilities can unlock Seedance 2.5 by paying US$0.99 or subscribing to any Wizstar membership plan, which also includes 30 seconds of complimentary video generation credits.
Wizstar will continue integrating leading AI models and enhancing its platform ecosystem to deliver a more open, efficient, and intelligent AI content creation experience for creators and enterprises worldwide.
For more information and the latest product updates, please visit https://wizstar.com/?channel_id=W_marketing_1
For more information about the new model releases and access benefits, please visit: https://wizstar.com/activitypages/seedance25?channel_id=W_marketing_1
** This press release is distributed by PR Newswire through automated distribution system, for which the client assumes full responsibility. **
Wizstar Launches Seedance 2.5 and MiniMax H3, Further Expanding Its AI Video Capabilities
SAN JOSE, Calif., Aug. 7, 2026 /PRNewswire/ -- At FMS 2026 on August 5, Longsys (301308.SZ) Chief Scientist Dr. Jian Chen delivered a keynote speech titled "Intelligence at the Edge: Powered by the Storage Foundry Model," addressing the new storage demands created as AI shifts from the cloud to the edge.
During the keynote, Dr. Chen introduced Longsys' Storage Foundry Model and presented three edge AI storage innovations — HLCache™, iSA™ + AISSD™, and AIDIMM™ — designed to address key industry challenges including high DRAM costs, limited memory capacity, and bandwidth constraints, creating a new pathway for edge AI storage innovation.
Engineering Culture: The Foundation of Storage Innovation
Dr. Chen traced Longsys' growth since its founding in 1999, including the launch of FORESEE, the acquisition of Lexar, expanded packaging and manufacturing through Longforce (Suzhou) Technology, and the integration of Brazil-based Zilia. Longsys now operates a global manufacturing network supporting worldwide delivery.
Rather than focusing solely on standardized products like traditional storage providers, Longsys applies full-stack engineering across eight areas — chip design, firmware development, material engineering, host software and OS-level adaptation, hardware design, packaging, automated testing, and mass production — enabling cross-domain hardware-software co-optimization and customized development. Together, these integrated capabilities across the storage value chain form the basis of the Storage Foundry Model.
Redefining Storage Collaboration for the Edge AI Era
Dr. Chen highlighted that while NAND Flash and DRAM have achieved significant cost reductions over decades, the rise of edge AI has introduced a new level of complexity. He described the fragmented edge AI landscape as a "thousand-handed challenge," where diverse architectures, AI models, device forms, and applications require coordinated solutions rather than standardized approaches.
From AI PCs and mobile devices to industrial systems and robotics, different scenarios require tailored trade-offs among performance, power efficiency, form factor, and security. The Storage Foundry Model goes beyond conventional storage offerings by enabling solutions tailored to customers' CPUs, SoCs, AI models, and system architectures — integrating hardware, firmware, and system-level optimization to accelerate development cycles and lower barriers to edge AI deployment.
Three Longsys Innovations Powering Edge AI
HLCache™ UFS leverages Longsys' self-developed High-Level Cache technology to intelligently manage hot and cold data between UFS and DRAM, reducing memory pressure and improving multitasking performance on mobile AI devices.
iSA™ + AISSD™ enables efficient local AI inference for AI PCs and AI agent hosts by offloading MoE parameters and KV cache to dedicated SSD resources, reducing DRAM usage while supporting large-scale AI models.
AIDIMM™ addresses memory bandwidth challenges in AI BOX and edge AI systems with a plug-in high-bandwidth memory architecture, featuring a native 256-bit interface and up to 307.2GB/s bandwidth. Its flexible design simplifies system upgrades and reduces hardware redesign efforts.
"Collabovate": Building the Future of Edge AI Together
Dr. Chen highlighted that as edge AI evolves from early adoption to large-scale deployment, diverse requirements across devices, architectures, and workloads cannot be addressed by a single product or technology.
To address this fragmented market, Dr. Chen introduced "Collabovate" — combining Collaboration and Innovation — as a new ecosystem collaboration approach. Through the Storage Foundry Model, Longsys is bringing together partners across chips, devices, and AI ecosystems to leverage complementary capabilities and accelerate customized edge AI solutions.
By opening its full-stack technology capabilities, Longsys aims to build a more collaborative storage ecosystem, enabling deeper industry collaboration and unlocking new possibilities for intelligent computing.
About Longsys
Founded in 1999, Longsys (301308.SZ) is a globally leading branded semiconductor memory enterprise, integrating R&D, design, packaging and testing, manufacturing, and sales services. For more information please visit https://www.longsys.com/, and follow Longsys on LinkedIn and Facebook.
** This press release is distributed by PR Newswire through automated distribution system, for which the client assumes full responsibility. **
FMS 2026 | Longsys Chief Scientist Jian Chen Highlights the Storage Foundry Model for Edge AI
FMS 2026 | Longsys Chief Scientist Jian Chen Highlights the Storage Foundry Model for Edge AI
FMS 2026 | Longsys Chief Scientist Jian Chen Highlights the Storage Foundry Model for Edge AI
FMS 2026 | Longsys Chief Scientist Jian Chen Highlights the Storage Foundry Model for Edge AI
