Shanghai AI Laboratory

Shanghai AI Laboratory (上海人工智能实验室, also known as Shanghai AI Lab) is a nonprofit artificial intelligence research institute headquartered in Shanghai, China, founded in July 2020 with backing from the Shanghai Municipal Government, the Chinese Ministry of Science and Technology, and a coalition of Chinese universities including Shanghai Jiao Tong University, Fudan University, Tongji University, and other partners. It develops the InternLM family of open-weights foundation models (InternLM, InternLM2, InternLM2.5, InternVL multimodal variants), the OpenCompass foundation-model evaluation framework, the OpenMMLab open-source computer vision toolbox ecosystem, and AI for science research output. As of April 2026, Shanghai AI Lab is one of the principal Chinese AI research bodies and is positioned as the Shanghai-based counterpart to Beijing-based BAAI within the broader Chinese AI research ecosystem.

At a glance

• Founded: July 2020 in Shanghai, China.
• Status: Nonprofit research institute. Backing from the Shanghai Municipal Government, the Chinese Ministry of Science and Technology, and a coalition of Chinese universities.
• Funding: Chinese government funding through the Shanghai Municipal Government and the Ministry of Science and Technology, plus contributions from member academic institutions. Specific budget allocations have not been publicly disclosed.
• Director: Yu Qiao, Director of Shanghai AI Laboratory and prominent computer-vision researcher. Previously Deputy Director of the Shenzhen Institute of Advanced Technology of the Chinese Academy of Sciences.
• Other notable leadership: Dahua Lin, Lead Scientist; computer-vision researcher and core contributor to the OpenMMLab ecosystem. Wenhai Wang, Lead Scientist on InternVL multimodal models. Jifeng Dai, Lead Scientist on InternImage and other computer-vision research.
• Open weights: Yes. The InternLM family is released open-weights through Hugging Face under permissive licensing (Apache-2.0 for selected variants). The OpenMMLab ecosystem and OpenCompass evaluation framework are open-source through GitHub.
• Flagship outputs: InternLM family (InternLM, InternLM2, InternLM2.5, InternLM2-Chat, InternLM-XComposer, InternVL, InternImage), OpenCompass (foundation-model evaluation framework), OpenMMLab (open-source computer vision toolbox ecosystem with MMDetection, MMSegmentation, MMPose, MMYOLO, and other projects), and AI for science research output.

Origins

Shanghai AI Laboratory was founded in July 2020 by the Shanghai Municipal Government in coordination with the Chinese Ministry of Science and Technology, with a research mandate to advance artificial intelligence research and serve as a coordinating body for the Shanghai-based AI research ecosystem. The founding member institutions included Shanghai Jiao Tong University, Fudan University, Tongji University, and other Shanghai universities. The Shanghai location reflected the Shanghai municipal government's 2018 to 2020 commitment to artificial intelligence as a strategic-priority research area, with the World Artificial Intelligence Conference (WAIC) hosted annually in Shanghai providing the policy context for the lab's establishment.

The 2020 to 2022 founding period built out research capacity across the principal research areas: foundation models, computer vision, multimodal AI, decision-making AI, and AI for science. Yu Qiao, the founding director, recruited senior research talent from the Chinese Academy of Sciences and other Chinese academic institutions. Dahua Lin, a computer-vision researcher and core contributor to the OpenMMLab ecosystem, anchored the computer-vision research direction.

The 2023 release of the InternLM family was Shanghai AI Lab's most consequential international research output. InternLM (initial release, 7B and 20B variants) provided a open-weights bilingual foundation model with Chinese and English capability. Subsequent InternLM2 (January 2024), InternLM2.5 (July 2024), and InternLM-XComposer (multimodal variant) iterations continued the open-weights foundation-model release cadence. The InternVL multimodal variants extended the family to vision-and-language capability.

The OpenCompass foundation-model evaluation framework, launched in 2023, has emerged as one of the principal Chinese-language foundation-model benchmark frameworks. The OpenMMLab open-source ecosystem, with precursor history in Dahua Lin's research at the Chinese University of Hong Kong before the Shanghai AI Lab founding, has continued to anchor the institute's open-research credibility. The OpenMMLab projects (MMDetection, MMSegmentation, MMPose, MMYOLO, MMOCR, and other) have continued to receive contributions from the broader Chinese and international computer-vision research community through 2024 to 2026.

The 2024 to 2026 period has continued the InternLM iteration alongside AI for science research output, including published research at Nature and other scientific journals. The continued World Artificial Intelligence Conference annually in Shanghai has provided the policy and commercial context for the lab's continued research output.

Mission and strategy

Shanghai AI Laboratory's stated mission is to advance fundamental artificial intelligence research and to serve as a coordinating body for the Shanghai-based AI research ecosystem, with emphasis on foundation models, multimodal AI, computer vision, and AI for science. The strategic premise reflects the Chinese government commitment to artificial intelligence as a strategic-priority research area, with Shanghai AI Lab explicitly positioned as the Shanghai-based counterpart to Beijing-based BAAI within the broader Chinese AI research ecosystem.

The strategy combines four threads. First, the InternLM open-weights foundation model family providing Chinese and English language capability and vision-language multimodal capability. Second, the OpenCompass foundation-model evaluation framework providing the principal Chinese-language foundation-model benchmark infrastructure. Third, the OpenMMLab open-source computer vision toolbox ecosystem providing open-source AI infrastructure for the broader research community. Fourth, AI for science research output spanning biological AI, weather forecasting, materials discovery, and other application domains.

The competitive premise is that the Shanghai-based AI research talent base, the coordinating role across Shanghai academic institutions, and the Chinese government research-funding commitment provide Shanghai AI Lab with a durable structural advantage as one of the principal Chinese AI research bodies.

Distribution channels include open-source distribution of the InternLM family through Hugging Face, open-source distribution of the OpenMMLab and OpenCompass ecosystems through GitHub, published research output through major academic venues, and cross-institution research-cooperation relationships across Chinese academic and industrial AI organizations.

Models and products

• InternLM family. InternLM (initial release, 7B and 20B variants), InternLM2 (January 2024), InternLM2.5 (July 2024), InternLM2-Chat (instruction-tuned variant), InternLM-XComposer (multimodal variant), InternVL (vision-language multimodal variants), InternImage (computer vision foundation model). Open-weights through Hugging Face under Apache-2.0 (selected variants) and other permissive licenses.
• OpenCompass. Foundation-model evaluation framework with Chinese-language benchmark suites. The principal Chinese-language foundation-model benchmark framework. Open-source through GitHub.
• OpenMMLab. Open-source computer vision toolbox ecosystem with precursor history at the Chinese University of Hong Kong. Major projects include MMDetection (object detection), MMSegmentation (semantic segmentation), MMPose (human pose estimation), MMYOLO (YOLO-family object detection), MMOCR (optical character recognition), MMTracking (multi-object tracking), MMAction2 (action recognition). Open-source through GitHub.
• AI for science research output. Published research at Nature and other scientific journals across biological AI, weather forecasting, materials discovery, and other application domains.

Distribution channels include open-source distribution through Hugging Face and GitHub, published research output through major academic venues, and cross-institution research-cooperation relationships.

Benchmarks and standing

Shanghai AI Laboratory's evaluation framework focuses on foundation-model benchmarks (with the OpenCompass benchmarking framework as the principal Chinese-language foundation-model evaluation infrastructure), computer-vision benchmarks through the OpenMMLab ecosystem, and AI for science benchmarks across biological AI, weather forecasting, and materials discovery. The InternLM family has been consistently characterized in Chinese-language NLP industry coverage as one of the principal open-weights bilingual foundation model families.

The OpenCompass benchmarking framework has continued to serve as the principal Chinese-language foundation-model evaluation infrastructure, with coverage across the major Chinese-language and bilingual foundation-model families (Alibaba Qwen, DeepSeek, Z.AI, Moonshot AI, Stepfun, MiniMax, Pangu, BAAI Aquila, and other families).

The OpenMMLab ecosystem has continued to receive contributions from the broader Chinese and international computer-vision research community through 2024 to 2026, with citation and adoption metrics across academic and industrial computer-vision research.

The AI for science research output has continued through 2024 to 2026 with published research at Nature and other scientific journals.

Leadership

As of April 2026, Shanghai AI Laboratory's senior leadership includes:

• Yu Qiao, Director of Shanghai AI Laboratory. Computer-vision researcher; previously Deputy Director of the Shenzhen Institute of Advanced Technology of the Chinese Academy of Sciences.
• Dahua Lin, Lead Scientist. Computer-vision researcher and core contributor to the OpenMMLab ecosystem. Precursor history at the Chinese University of Hong Kong.
• Wenhai Wang, Lead Scientist on InternVL multimodal models.
• Jifeng Dai, Lead Scientist on InternImage and other computer-vision research. Computer-vision researcher.
• Senior research leadership across the principal research areas (foundation models, computer vision, multimodal AI, decision-making AI, AI for science).

Departures and arrivals are continuous. The 2024 to 2026 period has continued senior research-talent recruitment from Shanghai academic institutions, the Chinese Academy of Sciences, and other Chinese research organizations.

Funding and backers

Shanghai AI Laboratory operates under Chinese government funding through the Shanghai Municipal Government and the Ministry of Science and Technology, plus contributions from the founding member academic institutions. Specific budget allocations have not been publicly disclosed, although industry coverage has placed Shanghai AI Lab's annual operating budget in the multi-hundred-million-yuan range comparable to BAAI.

The Shanghai government commitment to artificial intelligence as a strategic-priority research area provides Shanghai AI Lab with financial-runway certainty. Open questions on near-term resourcing are limited, given the multi-year research-mandate commitments.

Industry position

Shanghai AI Laboratory occupies a structurally distinctive position as one of the principal Chinese government-backed AI research bodies, with Chinese government backing, the InternLM open-weights foundation model family, the OpenCompass foundation-model evaluation framework, the OpenMMLab open-source computer vision toolbox ecosystem, and AI for science research output. The 2020 to 2026 research output and the open-research positioning have established Shanghai AI Lab as the Shanghai-based counterpart to Beijing-based BAAI within the broader Chinese AI research ecosystem.

Industry coverage has consistently characterized Shanghai AI Lab as one of the principal Chinese AI research bodies, alongside BAAI on the Beijing side, Huawei Noah's Ark Lab on the industrial-research side, and Tsinghua KEG, Tsinghua IIIS, and other Chinese academic AI research organizations on the university-research side. The 2024 to 2026 period has continued the lab's research output and the cross-institution research-coordination role.

Competitive landscape

• BAAI (Beijing Academy of AI). Chinese government-backed AI research peer with a similar coordinating role for the Beijing AI research ecosystem.
• Huawei Noah's Ark Lab. Chinese industrial AI research peer with overlap on foundation-model research.
• Z.AI (Zhipu AI), DeepSeek, Moonshot AI, Stepfun, MiniMax. Commercial Chinese AI startups with cross-institution research-cooperation relationships and talent flow.
• Tencent Hunyuan, Alibaba Qwen / DAMO, Baidu, ByteDance Seed. Chinese commercial AI peers with overlap on Chinese-language foundation model research.
• Tsinghua KEG, Tsinghua IIIS. Chinese academic AI research peers with cross-institution research-cooperation.
• Allen Institute for AI (Ai2), EleutherAI, LAION. Non-Chinese open-research AI peers with overlap on the open-source AI infrastructure positioning.
• Hugging Face. Open-research distribution partner; Shanghai AI Lab distributes the InternLM family through Hugging Face.

Outlook

• The continued cadence of InternLM family releases through 2026 and 2027.
• The continued OpenCompass benchmarking framework expansion as the principal Chinese-language foundation-model evaluation infrastructure.
• The continued OpenMMLab open-source ecosystem expansion and the cross-institution research-cooperation.
• The continued AI for science research output and the published research trajectory at Nature and other scientific journals.
• The competitive dynamic with BAAI and Huawei Noah's Ark Lab on Chinese government-backed AI research positioning.
• The continued senior research-talent recruitment and senior leadership stability through the 2026 to 2027 Chinese AI commercial expansion.

Sources

• Shanghai AI Laboratory official site. Institute reference (Chinese language).
• InternLM on GitHub. Open-weights foundation model family.
• OpenCompass. Foundation-model evaluation framework.
• OpenMMLab on GitHub. Open-source computer vision toolbox ecosystem.
• InternLM on Hugging Face. Open-weights model distribution.