DeepSeek planning to launch a new agentic GPT-5 rival as early as this year

Chinese artificial intelligence startup DeepSeek is gearing up to release a new AI system that could intensify competition in the global AI race. The Hangzhou-based company, which rose to prominence earlier this year with its rapid ascent in the sector, plans to unveil its most powerful model by the end of 2025, according to reports.

The upcoming system is expected to be an AI agent capable of autonomously executing multi-step tasks with minimal human input. Unlike traditional chatbots that require frequent prompting, agents are designed to operate with greater independence, carrying out complex workflows and streamlining productivity.

Sources familiar with the project suggest DeepSeek’s model will go a step further by incorporating adaptive learning, enabling it to learn from its past actions and refine its responses over time. This self-improving capability could set it apart from conventional, static AI systems and enhance its long-term usability in real-world scenarios.

Industry analysts say such features highlight a broader trend in the AI sector, where companies including OpenAI, Google, Anthropic, and Meta are investing heavily in autonomous, agent-driven technologies. Businesses and professionals are increasingly looking for AI systems that can manage repetitive, time-consuming tasks without constant oversight.

DeepSeek’s entry into this space also reflects China’s wider push to strengthen its position in advanced technologies and compete directly with leading US firms. If the company succeeds in delivering a more efficient and potentially less resource-hungry agent than existing models, it could disrupt the market once again, echoing the impact of its earlier breakthroughs.

For now, details about the system remain limited, but with a launch targeted for late 2025, Silicon Valley and the global AI industry will be closely watching how DeepSeek positions itself against giants like OpenAI, Google, and Microsoft.