DeepSeek develops AI agent model to challenge OpenAI: sources

DeepSeek, a Hangzhou-based AI startup, is developing a new AI model with advanced agent features to compete with US firms like OpenAI, according to sources familiar with the project.

The founder is pushing for an unveiling of the model in Q4 2025, and it is designed to carry out complex, multi-step actions for users with minimal input and learn from its past actions.

DeepSeek previously released an AI model called R1 in January, which drew industry attention for its reasoning abilities and benchmark performance.

Since then, the company has released only minor updates, while competitors in China and the US have launched new models and features.

The company’s plan for a new agent-focused model mirrors a broader trend in the industry toward developing AI agents capable of handling more sophisticated tasks beyond basic chatbot responses.

The company has not commented on the project or its expected release timing.

🔗 Source: Bloomberg

🧠 Food for thought

Implications, context, and why it matters.

AI agents represent the industry’s next competitive battleground

DeepSeek’s agent development aligns with a broader industry shift toward more autonomous AI systems that can perform multi-step tasks with minimal human oversight.
OpenAI and Microsoft have also launched their own agent software in recent months, demonstrating that major players view agents as the next frontier beyond traditional chatbots1.
This industry-wide push reflects the recognition that simple text responses are insufficient for complex real-world applications, with companies racing to build systems that can research vacations, write code, and handle sophisticated workflows autonomously.

DeepSeek faces intensified competition from well-funded Chinese rivals

While DeepSeek achieved remarkable efficiency with its R1 model—costing just several million dollars to build while matching OpenAI’s performance—Chinese competitors have dramatically increased their investment pace1.
Alibaba pledged 380 billion RMB ($52.9 billion) over three years to enhance AI and cloud infrastructure, demonstrating the scale of resources being deployed against DeepSeek2.
DeepSeek’s “muted pace of development” compared to rivals like Alibaba and Tencent, who maintain “frenetic” rollout schedules, suggests the company may struggle to maintain its early momentum without matching their resource-intensive approach1.