Chengdu Mingtu Technology Co., Ltd. (成都明途科技有限公司) is a privately held AI company registered in Chengdu, Sichuan, with an establishment date of 11 July 2014 and legal representative Xiao Xuesong (肖雪松). Business-registry aggregators list its registered capital as RMB 30 million and paid-in capital as RMB 14.335 million, with an insured employee count in the mid-double digits (as of 24 January 2026), but those capital and headcount figures should be treated as database-reported registry snapshots rather than audited disclosures. Mingtu appears on the publicly released list for the sixth batch of national Specialized and New “Little Giant” enterprises (公示名单 dated September 2024), while other labels often attached to the firm such as “national high-tech enterprise” and “gazelle enterprise” are widely repeated in company profiles and media features but are not consistently traceable to a single authoritative public register in the same way and should therefore be treated as reported claims unless independently confirmed from the issuing authority.
Public reporting and company event coverage describe Mingtu’s current positioning around a vertical “work” large model branded WorkBrain and related digital-human products aimed at government and enterprise use cases, with a timeline that includes the May 31, 2023 launch of a dialogue-style “work digital human” platform branded WorkChat, a November 2023 release event for WorkBrain V3.0, and a March 2025 event for WorkBrain V5.0 held at Chengdu’s Kechuang Ecological Island. Descriptions of WorkBrain’s architecture and performance, including parameter scale and acceleration figures, are primarily presented through partner ecosystem announcements and company-facing narratives rather than through independently reproducible evaluations, so they should be read as vendor claims even when repeated by major platform partners. In July 2024, Huawei’s Ascend ecosystem communications described WorkBrain as having completed compatibility testing and receiving an Ascend “native” technical certification and characterized Mingtu as building on Ascend hardware and MindSpore tooling, while the same materials presented pilot or rollout examples in entities such as State Grid and local cultural-tourism deployments, which should be understood as partner-and-company described implementations rather than fully verifiable enterprise-wide adoptions.
For regulatory status, credible local and national reporting in late 2024 described Mingtu as having achieved “double filing” (双备案) in China’s AI governance context, referring to a deep-synthesis algorithm filing for a work-content generation algorithm and a generative-AI service registration for a model branded “MT-WorkGPT,” but the specific product naming, scope of service, and operational deployment implied by these filings should be kept distinct from broader marketing statements about capabilities. Mingtu’s product family is commonly presented as a suite spanning a digital-human cloud platform (WorkChat), an “embodied” or device-linked interaction layer often branded WorkBox, a goal/decision workflow layer branded WorkDI, and industry “digital brain” offerings branded WorkChain; claims about nationwide footprints such as “three R&D centers,” “five marketing regions,” “seven branches,” “800+ partners,” and “2,000+ government and enterprise clients,” as well as large-scale interaction totals and influencer-related counts, appear in company and promotional profiles and are not readily verifiable through primary filings, so they should be treated as company claims rather than stated as confirmed facts. Within the digital-human scope, Mingtu’s offerings are best characterized, based on higher-quality reporting and partner announcements, as AI-driven virtual human interfaces for role-based interaction over organizational knowledge bases and workflows, typically positioned for private-data deployment and task-oriented service in domains such as public-service consultation, enterprise operations support, venue or exhibition guidance, and sectoral information services, while more expansive assertions about universal “digital twins” of real individuals, routine legal or medical advising, or quantified productivity impacts should be treated as marketing narratives unless corroborated by audited case studies, regulated disclosures, or clearly attributable executive statements tied to specific deployments and dates.