UNISOUND (09678) released its interim results, with revenue increasing by 38.7% year-on-year to 562 million yuan, and the commercialization capability of large model technology further enhanced.

UNISOUND (09678) released its interim results for the six months ending June 30, 2026, reporting a revenue of 562 million yuan (RMB, the same below), representing a year-on-year increase of 38.7%; gross profit of 186 million yuan, an increase of 42% year-on-year; and a loss attributable to shareholders of 237 million yuan, which narrowed by 20.2% year-on-year. The announcement stated that the group's total revenue increased by 38.7% year-on-year, driven by enhanced commercialization capabilities of large model technology. Supported by continuous iterative upgrades of the native intelligent agent large model technology and the groups experience in providing intelligent services across various verticals, the groups enterprise intelligent service revenue segment grew steadily, while the model invocation Token business experienced explosive growth. In response to the paradigm shift in the industry, the company persists with its "strong foundational model, deep application" strategy, focusing on coordinated efforts across three layers: foundational capability layer, intelligent construction layer, and value operation layer. At the foundational capability layer, during the reporting period, the company launched the native intelligent agent large model U2, leveraging a computing cluster and high-quality business data-driven post-training to achieve stronger intelligence with lower resource consumption, realizing "high intelligence density, high Token value," and turning clients' business objects, processes, rules, and actions into reusable assets. At the intelligent construction layer, the intelligent agent platform UniAgent OS encapsulates model capabilities into action-capable agents, while the business operation platform UniOps abstracts client operations into a runnable business map, ensuring that agent actions are well-founded. At the value operation layer, the company implements a closed-loop of planning decision-making execution feedback for every delivery, where key steps involve joint decision-making with humans, crucial nodes are confirmed by humans, and execution results are collected in real-time, accumulating experience and continuously optimizing, completing the last mile of value realization and providing clients with measurable increases in revenue, cost reductions, and efficiency improvements. From an operational perspective, the company achieved a revenue of 562 million yuan during the reporting period, representing a year-on-year growth of 38.7%. Among this, the enterprise intelligent service generated a revenue of 478 million yuan, a 35.7% year-on-year increase; the large model invocation Token business started from virtually zero, achieving nearly 30 million yuan in revenue, with strong growth momentum; the edge-side AI business showed steady growth, generating nearly 53.8 million yuan in revenue. Behind the revenue growth is the simultaneous improvement in platform capabilities and asset reuse levels, as the company is transforming AI capabilities into measurable operational value for clients along three dimensions: increasing revenue, reducing costs, and enhancing efficiency.