Original · Unique Research · 2026-07-16
Editor's note: The first-person report and its judgments belong to the original Chinese author. This English rendition retains all statistics, award categories, judging dimensions, and organizational descriptions from the source. All percentages and figures are sourced from the awards application pool and are self-reported by applicants, not independently verified.
Unique Awards
The 7th Unique Awards 2026 Announced: Benchmarks of Global AI Business Innovation
"The application layer is the main battlefield, globalization is the default, multi-model and hybrid deployment are the engineering norm, and cost reduction and efficiency gains are numbers you can actually put on the table."
In this year's Unique Awards applications, nearly 80% of products are rooted in the application layer; nearly 70% reach overseas markets (cross-border, going global, or Hong Kong/Macau/Taiwan); and products connecting three or more large models are the mainstream, not the exception. Globalization and multi-model routing have shifted from "bonus points" to "default options."
On July 16, 2026, at the 2026 Unique Grand Awards · Shanghai AI Business Summit, the 7th Unique Awards were officially announced.
Key Observations from This Year's Cohort
We dissected every field in this year's Unique Awards product and case application forms. Several signals are clear.
The application layer remains the absolute main battlefield. Among AI-native products, 76.1% fall in the application layer, 18.8% in the middle layer, and only 5.1% in the hardware layer. On the business case side, the application layer share rises further to 82.4% — 6.3 percentage points higher than the product side — with the middle layer at 13.7% and the model layer at just 3.9%. The closer to real business scenarios, the higher the concentration at the application layer. The most contested ground in this year's AI competition is still the layer closest to users and scenarios.
Globalization is already the default configuration. Only 30.8% of products serve purely the domestic China market with no overseas or cross-border business. In other words, nearly 70% (69.2%) of applicants already cover cross-border, Hong Kong/Macau/Taiwan, or overseas markets. Among them, 50.4% directly reach target overseas markets (the U.S., Southeast Asia, Europe, Japan/Korea, the Middle East, Latin America, Africa) — meaning one in two products has truly "gone out to sea." By region: China cross-border market coverage is 47.9%, the U.S. 39.3%, Southeast Asia 38.5%, Hong Kong/Macau/Taiwan 35.9%, Europe 31.6%, Japan/Korea 29.1%, Latin America 21.4%, the Middle East 19.7%, and Africa 13.7%. The coverage is already very broad, not concentrated on one or two markets.
Few bet on a single model. Only 18.8% of products connect to a single large model; the remaining 81.2% use multi-model access. Products connecting to three or more models account for 70.1%, with an average of 6.2 models per product. Model choices are also diversified: Alibaba Qwen 58.1%, ChatGPT 50.4%, DeepSeek 47.9%, ByteDance Doubao 44.4%, Gemini 41.9%, Zhipu GLM 34.2%, Claude 29.1%. No single model holds an absolute majority share; multi-model routing is an engineering standard, not a technical showcase.
Deployment: cloud and private deployment go hand in hand. Public cloud deployment at 86.3% remains mainstream, but 35.9% also have private cloud deployment capability, 36.8% support on-premises deployment, and 33.3% offer customer private deployment — all three exceed one-third. This means this year's products are generally not "Demos that only run in the cloud" but deliverable systems that can adapt to enterprise-grade, compliance, and data-sensitive scenarios. Among cloud vendors, Alibaba Cloud leads at 67.9%, followed by Tencent Cloud 37.5%, Volcano Engine 33.0%, and Amazon Web Services 24.1%. On the chip side, NVIDIA accounts for 67.6% and Huawei Ascend 39.7% — domestic chip adaptation is substantial.
Technical native strength is concentrated in Agent capabilities. On the product side: prompt engineering 79.5%, workflow orchestration 70.9%, model tool calling 62.4%, retrieval-augmented generation 59.8%, multi-agent collaboration 59.0%, Agent frameworks 55.6%. On the case side: workflow orchestration 72.5%, prompt engineering 72.5%, retrieval-augmented generation 62.7%, multi-agent collaboration 56.9%, fine-tuning/supervised fine-tuning 51.0%. The fine-tuning rate on the case side is 14.2 percentage points higher than the product side's 36.8%, showing that business cases are more willing to do deep customization for specific clients rather than directly applying generic capabilities. "Agents truly entering business closed loops" has shifted from a summit theme to a verifiable technical fact on the application forms.
Products are uniformly "new," but team sizes have diverged. Products launched within the last 12 months (July 2025 to present) account for 70.9%, products launched in 2024 or later account for 88.9%, and "old products" launched in 2022 or earlier account for only 3.4%. This year is almost entirely a cohort of new products competing on the same stage. But team sizes have clearly diverged: teams under 20 people account for 36.8% (of which under 10 people account for 22.2%), with a median team size of 30. Meanwhile, teams over 100 people account for 22.2% and teams over 300 people account for 11.1% — small teams iterating quickly and large teams scaling delivery, both approaches coexisting.
Customer coverage breadth is widening, and commercial scale is already substantial. 67.0% of products simultaneously serve three or more enterprise-size segments (from micro to ultra-large), 19.8% cover all five size segments simultaneously, and only 16.5% focus on a single segment. Pure 2B products account for 40.2%, while products simultaneously covering 2B, 2C, or 2ProC customer groups account for 41.9%. Commercial scale on the case side is equally impressive: cases with budgets in the 1 million–5 million RMB range account for the largest share at 31.4%, cases with budgets above 5 million RMB account for 17.6%, and those above 10 million RMB account for 7.8%. Cases where AI-related spending can be separately quantified and tracked (rather than lumped into project totals) account for 37.3%, showing that AI investment is shifting from a "hidden cost" to a separately accountable budget line item.
Cost reduction and efficiency gains are numbers that can be put on the table. Among submitted business cases, reduced operational costs are verified in 68.6%, reduced labor costs in 64.7%, shortened business processing time in 62.7%, improved sales conversion rates in 51.0%, improved customer acquisition efficiency in 41.2%, improved content/creative output efficiency in 39.2%, and improved customer satisfaction in 37.3%. Over 60% of cases can produce quantitative cost-reduction data, speaking with real business results.
Founder generations are stretching at both ends. Founders born in the 1990s or later account for 43.0% of applicants, approaching half; 1980s-born founders account for 40.4%, still the main cohort. Meanwhile, senior founders born before 1980 with years of industry experience account for 16.7%, of which those over 50 years old account for 7.9%. Young teams bringing new technology into traditional sectors and senior practitioners with accumulated industry expertise re-entering the field — this is the most interesting demographic structure in this year's applicant pool, echoing the two types of people the Unique Awards has always sought: technology-native newcomers and industry-native veterans.
Award Winners
AI-Native Product Award
This image may have used AI-generated technology; please exercise caution in identifying it.
AI Business Case Award
Award Categories
The 7th Unique Awards established two award categories covering the core forms of AI innovation from product to implementation.
AI-Native Product Award: For standardized software and hardware products and applications built on AI technology, including AI-native software, Agent applications, vertical AI tools, enterprise-grade Agent platforms, AI hardware, model services, and AI infrastructure. Review focuses on product technical architecture innovation, commercial scalability, user experience and scenario fit, as well as technical native strength and differentiated moats.
AI Business Case Award: For vertical-domain solutions customized for enterprises using AI technology, focusing on implementation effects, business value, and replicable experience in real customer scenarios. Review focuses on implementation results and business growth data, industry empowerment value and replicability, scenario innovation depth and problem-solving depth, and the commercial closed-loop capability from POC to scaled deployment.
Judging Dimensions
This year's judging revolved around four dimensions:
Product New-Quality Strength (Product): Evaluating product uniqueness, user experience, market foresight, and scalable potential
Technical Native Strength (Technology): Measuring underlying technical innovation, AI-native architecture, and engineering implementation capability
Commercial Implementation (Economic): Verifying real customers, growth data, business models, and scaling potential
Scenario Innovation (Scenario): Evaluating the depth of integration and innovative value between the product or case and specific business scenarios
Judging uses a weight-allocation mechanism, prioritizing product innovation and technical breakthroughs while validating real implementation value through business and scenario metrics, combining quantitative data, application materials, customer cases, and expert review to form a multi-dimensional judgment.
About the Unique Awards
The Unique Awards is a global AI business innovation benchmark award initiated by Unique Research (非凡产研), held for seven consecutive editions since its founding in 2019. As China's first professional award focused on AI business implementation practice, the Unique Awards is dedicated to discovering generative AI's native products, innovative scenarios, and benchmark cases worldwide, covering three dimensions: deep domestic cultivation, going global expansion, and global layout, promoting deep AI empowerment and value reconstruction across eight major business scenarios including marketing, content, sales, and experience.
The award defines digital-intelligent business standards with a forward-looking perspective. Past winning cases have covered 20+ fields including e-commerce, finance, and manufacturing. It also builds a four-dimensional judging system of "product new-quality strength + technical native strength + commercial implementation + scenario innovation," jointly analyzing case methodologies with top investment institutions and industry think tanks, and has collected 400+ practical cases, providing practitioners with a reusable toolkit for digital-intelligent transformation.
About Unique Research
Unique Research is an authoritative third-party organization focused on the AI field. Based on open-source and neutral principles, it publishes rankings of global AI company revenue and reach, with reproducible methodologies and data. Its findings are cited by top investment and academic institutions, providing reliable data benchmarks and decision-making references for investors and entrepreneurs, and driving industry innovation.
The Unique Grand Awards initiated by Unique Research is an industry event platform deeply tied to the Unique Awards. Since its founding in 2019, it has successfully held multiple annual summits and is a bellwether in the AI business field. Each edition attracts deep participation from global technology companies, AI unicorns, industry leaders, and top investment institutions. In the future, as AI advances toward general artificial intelligence, Unique Research will continue to focus on the commercial implementation and practice of AI applications, becoming a key hub connecting technology, business, and policy.
"From 2019 to 2026, the Unique Awards has always looked at the same thing: has AI truly entered the business, rather than remaining stuck in concepts? This year's data has already given the direction — the application layer is the main battlefield, globalization is the default, multi-model and hybrid deployment are the engineering norm, and cost reduction and efficiency gains are numbers you can actually put on the table."