TLDR: Chinese AI startup DeepSeek is preparing to launch a sophisticated AI agent by the fourth quarter of 2025, designed to perform complex, multi-step tasks with minimal user input and learn from its past actions. This strategic move intensifies its competition with leading U.S. AI firms like OpenAI, Anthropic, and Microsoft, which have already introduced similar agentic software. DeepSeek, known for its cost-efficient R1 model, seeks to reassert its innovation in the rapidly evolving AI landscape.
Chinese artificial intelligence firm DeepSeek is reportedly on track to release a highly advanced AI agent by the close of 2025, a development poised to significantly escalate the global rivalry in artificial intelligence. According to a Bloomberg report and other media outlets on Thursday, September 4, 2025, the Hangzhou-based startup is developing an AI model capable of executing multi-step tasks on behalf of users with remarkably minimal direction. The system is also engineered to continuously learn and refine its performance based on prior interactions and actions.
DeepSeek founder, Liang Wenfeng, is reportedly driving his team towards a launch in the final quarter of this year, signaling an ambitious push to compete directly with established U.S. giants such as OpenAI. This move comes as the AI industry increasingly shifts towards creating more autonomous agents that can handle intricate professional and personal tasks without extensive human supervision.
The company gained international attention in January with the release of its R1 model, a large language model (LLM) that demonstrated impressive capabilities in emulating human reasoning. Notably, the R1 model was developed at a reported cost of less than $6 million, a fraction of the billions typically invested by Western AI labs. This cost-efficiency has been a key differentiator for DeepSeek, enabling it to offer competitive token pricing, reportedly between $0.10-$0.20.
However, since the initial buzz around R1, DeepSeek has primarily released minor updates, while its competitors in both the United States and China have rapidly rolled out numerous new models and agent software. Major players like OpenAI, Anthropic, and Microsoft have already introduced their own agent-style platforms in recent months, aiming to make AI more autonomous and task-driven. China-founded Manus AI is another notable contender in the general-purpose AI agent space.
Also Read:
- UBS Forecasts Major AI Agent Monetization Boom in China by 2026
- China’s Open-Source AI Models Rapidly Closing Gap with Global Proprietary Leaders
DeepSeek’s latest endeavor is seen as a crucial step to re-establish its innovative edge. The company also recently launched its updated DeepSeek-V3.1 AI model last month, which reportedly surpasses its R1 model on key benchmarks and is optimized for compatibility with Chinese-made chips. This focus on domestic chip compatibility underscores the escalating tech tensions and China’s strategic objective to reduce reliance on American-made technology in the global AI race. DeepSeek’s advancements have already attracted a significant user base, with 125 million monthly active users globally by Q2 2025, and the firm secured $520 million in Series C funding, valuing it at $3.4 billion post-money.
