Shenzhen Launches Operational Guidelines for AI Pioneer City Project Support Plan

AIbase基地

Published inAI News · 4 min read · Mar 20, 2025

The Shenzhen Municipal Bureau of Industry and Information Technology has officially issued the "Implementation Rules for the Pioneer City Project Supporting Plan for Artificial Intelligence," aimed at promoting the high-quality development of Shenzhen's AI industry. This plan focuses on computing power support, fundamental data, AI software and services, encompassing multiple support programs to foster the AI ecosystem.

Large Model

Image Source Note: Image generated by AI, licensed by Midjourney.

A highlight is the "Model Voucher" program. This supports enterprises in using generative AI models registered with the Cyberspace Administration of China for model services and intelligent agent development. Enterprises can apply for "Model Vouchers" for subsidies, with a maximum annual application limit of 2 million RMB per enterprise. 30% of model service fees can be offset using these vouchers.

Beyond the "Model Voucher" program, Shenzhen will also fund demonstration and flagship application projects. Eligible demonstration projects can receive up to 2 million RMB in support, while flagship projects can receive up to 10 million RMB. Shenzhen will also provide matching support (1:1 ratio) for units undertaking national AI "challenge-seeking" projects, up to a maximum of 10 million RMB.

Support is also available for projects fostering domestic AI ecosystem innovation centers. Eligible projects will receive funding, with up to 30% subsidy for model application migration and adaptation services, capped at 1.5 million RMB per enterprise. Open-source AI software projects will also receive tiered funding.

For high-end exhibitions, forums, and competitions, Shenzhen will provide subsidies of up to 50% of the approved total investment, with a maximum of 3 million RMB. Support measures for embodied intelligent robot R&D and promotion will also be implemented, with companies receiving up to 3 million RMB in rewards for annual blockbuster products achieving certain sales volumes.

These rules will be in effect from March 18, 2025, to June 30, 2027. Shenzhen aims to further drive AI industry development and enhance the city's technological competitiveness through these measures.

Key Highlights:
🌟 Shenzhen launches an AI Pioneer City Project support plan, focusing on high-quality industry development.
💰 The "Model Voucher" program provides subsidies for enterprise model service fees, with a maximum annual application limit of 2 million RMB.
🏆 Up to 10 million RMB in funding for demonstration and flagship application projects, supporting AI ecosystem innovation.

Midjourney Core Developer theseriousadult Departs, Joins Cursor to Develop AI Programming Agents

theseriousadult, the key developer behind Midjourney's V2 through V7 models, announced their departure from the company on X. Over three years at Midjourney, this developer spearheaded multiple model iterations, renowned for their focus on aesthetics and creativity, and their work has received widespread industry acclaim. Following their departure, theseriousadult announced joining Cursor to develop AI programming agents. This new role...

High School Student Creates AI Model Evaluation Website Using Minecraft

In today's rapidly advancing AI landscape, effectively evaluating and comparing the capabilities of different generative AI models is a significant challenge. Traditional AI benchmarking methods are increasingly showing their limitations, prompting AI developers to explore more innovative evaluation approaches. Recently, a website called "Minecraft Benchmark" (MC-Bench) has emerged, uniquely leveraging Microsoft's sandbox game Minecraft to facilitate model assessment.

Google AI Studio Image Generation Feature Upgrade: Reduced Safety Misclassifications and Improved Usability

Google recently announced a major upgrade to the image generation capabilities of its AI development platform, Google AI Studio. This update significantly reduces the rate of safety misclassifications while significantly improving usability, providing users with a more efficient and accurate AI generation experience. Google AI Studio, a generative AI platform for developers, supports the integration of multimodal models such as text and images. This optimization of the image generation feature addresses previous issues caused by overly strict safety filters that led to misclassifications. The new version...

Chinese-made Miniature Deep-Sea Robot Achieves Multi-modal Movement at 10,000-meter Depth, Published in Top International Academic Journal

Recently, a miniature deep-sea multi-modal robot jointly developed by Beihang University, the Institute of Deep-Sea Science and Engineering of the Chinese Academy of Sciences, and Zhejiang University, was successfully published in the top international academic journal Science Robotics. This robot has milestone significance in ocean exploration and scientific research, particularly in deep-sea applications. Measuring less than 50 centimeters long and weighing only 1500 grams, this remarkable robot can perform multi-modal movement at a depth of 10,000 meters. This means it can not only swim, but also glide and crawl, demonstrating exceptional capabilities.