Wenzhun Intelligence (稳准智能), the Chinese company behind the LimiX structured-data model, has closed back-to-back Series A and A+ rounds. Chinese media reported on November 16, 2026, that the two rounds together are worth several hundred million yuan.

The exact amount was not disclosed. “Several hundred million yuan” means at least RMB 100 million, which is about $14 million.

Guoke Investment and Heyu Capital co-led the Series A. CDH Investments, Shanghai Guotou Xiandao and Xuhui Science and Technology Investment backed the A+ round. Gaohu Capital was the exclusive financial adviser.

Wenzhun Intelligence funding: Series A and A+ details

The company is registered in the Xiong’an New Area in Hebei and also has operations in Beijing. It was set up in October 2024 by a team from Tsinghua University’s computer science department, led by Professor Cui Peng.

DetailInformation
RoundsSeries A and Series A+
Combined amountSeveral hundred million yuan (exact figure not disclosed)
Series A co-leadsGuoke Investment, Heyu Capital
Series A+ investorsCDH Investments, Shanghai Guotou Xiandao, Xuhui Science and Technology Investment
Financial adviserGaohu Capital
ValuationNot disclosed
RegisteredXiong’an New Area, Hebei, China
FoundedOctober 2024

The new money will go into improving the core algorithms for its large data model, building models for specific industries, and commercial deployment, according to Sina.

Wenzhun Intelligence funding history

The A and A+ rounds bring the company to four rounds since it was founded. The first check came from the Xiong’an Science and Technology Innovation Angel Fund, which made Wenzhun its first investment.

According to PEdaily (投资界), the angel deal went from first meeting to closing in two months. The same report says the company’s valuation has multiplied several times in two years.

RoundAmountInvestors
AngelRMB 20 million (about $2.8 million)Xiong’an Science and Technology Innovation Angel Fund
Pre-ANot disclosed separatelyOriental Fortune Capital (东方富海), among others
Series APart of several hundred million yuanGuoke Investment and Heyu Capital (co-leads)
Series A+Part of several hundred million yuanCDH Investments, Shanghai Guotou Xiandao, Xuhui Science and Technology Investment

What is LimiX, the structured-data model behind Wenzhun Intelligence

Most AI funding goes to large language models, which work with text. LimiX (极数) is built for structured data: tables, time series and other numeric records found in factories, power grids, labs and banks.

The company calls it a large data model (LDM). It says LimiX was China’s first general-purpose model of this kind.

Wenzhun and Cui Peng’s Tsinghua team open-sourced the first two LimiX versions, with 2 million and 16 million parameters, on August 29, 2025. The LimiX technical report says one model can handle classification, regression, filling in missing values and data generation. It does not need separate training for each task.

LimiX-2 benchmark results

On September 16, 2026, the company released LimiX-2 with 400 million parameters. It reported overall Elo scores of 1,935 on TabArena, 1,432 on BCCO and 1,506 on TALENT, three public benchmarks for structured-data models.

The company says LimiX-2 ranked first on all three, ahead of comparable models from Google, SAP and Amazon. Those rankings come from the company’s own announcement, as reported by Stockstar.

How a data model differs from a language model

Wenzhun vice president Yu Fei compared the two in a June 2026 interview with The Paper. He described language models as good at writing, and LimiX as good at math and logic.

The model takes in industrial, financial or biological data and looks for cause-and-effect links between variables. It then runs prediction, classification, regression, anomaly detection and causal inference.

Its closest foreign rivals are tabular foundation models such as TabPFN from Germany’s Prior Labs and Amazon’s Mitra.

Where LimiX is used

The company says it has worked on more than 150 deployment scenarios. In February 2026, Cui Peng told Shanghai Securities News that LimiX had been tested in more than 20 industries.

One user is an electricity trading company that uses the model to forecast demand spikes during extreme weather and holidays. Another is a metals research center in Shenzhen, which uses it to link element ratios to material performance and narrow down alloy recipes.

China AI funding context in November 2026

Large AI rounds in China have continued this quarter. Kuaishou’s Kling AI unit completed a financing of about RMB 19.8 billion ($2.91 billion), TechNode Global reported on November 2. Video model startup Sand.ai raised $200 million in early November.

Funding in Asia was also active in October. Manus AI raised more than $500 million, according to this weekly AI funding report for October 7 to 12.

Wenzhun Intelligence is in a narrower category: models for the numeric data that companies already store. For deals and acquisitions elsewhere in the AI stack, see the roundup covering AMD’s World Labs purchase.

FAQ

How much did Wenzhun Intelligence raise?

Wenzhun Intelligence raised several hundred million yuan across consecutive Series A and A+ rounds, reported on November 16, 2026. The exact amount was not disclosed.

Who invested in Wenzhun Intelligence?

Guoke Investment and Heyu Capital co-led the Series A. CDH Investments, Shanghai Guotou Xiandao and Xuhui Science and Technology Investment invested in the A+ round.

What is LimiX?

LimiX is a large data model for structured data such as tables and time series. Wenzhun Intelligence and Tsinghua University developed it and open-sourced it in August 2025.

How large is LimiX-2?

LimiX-2, released on September 16, 2026, has 400 million parameters. The first open-source LimiX versions had 2 million and 16 million parameters.

Where is Wenzhun Intelligence based?

Wenzhun Intelligence is registered in the Xiong’an New Area in Hebei, China, with operations in Beijing. It was founded in October 2024.

Sheldon researches company ownership and business models for AIFundingtracker.com, covering who controls tech and AI companies and how they make money. He also writes plain-language guides to online platforms and what readers should know before using them.

Leave A Reply