Chinese Power Reshapes AI Hegemony! DeepSeek Crowned World's Second-Largest AI Lab, Open-Source Ecosystem Reaches Milestone

DeepSeek’s leap is a milestone for open-source AI!” — This comment by an overseas user on platform X garnered over 300,000 views in one day, capturing the global AI community’s astonishment at China’s dark horse.

1 .Claiming #2 Globally: DeepSeek Rewrites AI Power Dynamics

On May 30, 2025, independent evaluator Artificial Analysis released its latest global AI lab ranking: Chinese team DeepSeek-R1-0528 achieved a comprehensive score of 68 points, surpassing Elon Musk’s xAI (Grok-3 mini), Mark Zuckerberg’s Meta (Llama 4), and Anthropic (Claude series) to tie with Google Gemini 2.5 Pro as the world’s second-ranked AI lab, trailing only OpenAI’s o4-mini (70 points).

This marks the first time a Chinese AI model has entered the global elite tier in an authoritative index. The Artificial Analysis Intelligence Index covers seven rigorous tests—including mathematical reasoning, code generation, and scientific thinking—recognized as the “gold standard” for measuring AI capability.

Critically, DeepSeek is the only open-source model in the top three, shattering the “closed-source monopoly on high-end AI.”

2. Technical Breakthrough: Small Parameters, Big Wisdom – Dominating Cost and Efficiency

This achievement did not rely on traditional “parameter scaling” but instead leveraged reinforcement learning (RL) to optimize inference architecture, enabling an “intelligence leap”:

Transformative gains in mathematical ability: Accuracy on the AIME 2025 math competition test surged from 70% to 87.5%, rivaling human Olympiad contestants;

in open-source code generation: Scored 59 points on LiveCodeBench, nearing closed-source leader Claude 4 Opus;

Hallucination rate reduced by 50%, enhancing reliability with complex documents, plus support for 128K context + tool calling.

More striking is its cost efficiency: Inference costs just $0.96 per million tokens—under one-third of OpenAI’s o4-mini—with total training costs below $6 million.

3. User Explosion: Monthly Active Users Surpass ChatGPT, World’s #3 AI Tool

Technical strength directly fueled market momentum. In February 2025, DeepSeek reached 525 million monthly visits, surpassing ChatGPT (500 million) for the first time. Its market share surged to 6.58%, making it the world’s third-largest AI tool after ChatGPT (43.16%) and design platform Canva (8.27%).

Its growth curve is “phenomenal”:

After launching the R1 model in January 2025, monthly users exploded by 125 million, with 80% growth occurring in the final week;

Official app downloads exceeded 100 million in one month, with daily active users topping 30 million;

In the chatbot segment, it secured 12.12% market share, ranking #2 globally.

4. Open-Source Revolution: China Powers 40% of Global Ecosystem, China-US Gap Narrows to 0.3%

Behind DeepSeek’s rise lies China’s collective open-source momentum. According to Stanford’s 2025 AI Index Report:

Chinese models now contribute over 40% to the global open-source ecosystem;

DeepSeek and Alibaba’s Tongyi Qianwen form a dual-engine strategy—the former excelling in depth, the latter in breadth;

The China-US AI capability gap narrowed from 17.5% in 2023 to just 0.3%.

Released under the MIT License, DeepSeek-R1 allows free commercial use, driving inclusive tech access. The team distilled its core technology into an 8B-parameter compact model that matches Alibaba’s 235B-parameter closed model in math benchmarks—paving the way for low-cost AI deployment in SMEs.

5. The Next Battle: R2 Multimodal Model Launching Within Six Months, Challenging OpenAI’s Throne

DeepSeek has announced: The R2 model will debut by end-2025, focusing on multimodal breakthroughs (image + voice + text fusion) and targeting OpenAI’s GPT-5.

Industry observers predict: “If R1 is only 2 points behind OpenAI, R2 may overtake it.”

X users debate: *“How long can GPT-5 stay ahead?”*

While Silicon Valley giants refine “black-box models” behind closed doors, a Chinese company delivers elite AI to millions of developers through open-source, efficiency, and democratization.

DeepSeek’s rise isn’t just a technical victory—it’s a dimensional reduction strike by open ecosystems against closed hegemony.

As its founder stated: “When technical barriers fall, humanity’s path to AGI will accelerate beyond all imagination.”