DeepSeek agent platform reaches 1.54 million daily active users

Rapid adoption of Harness

DeepSeek, the Hangzhou-based startup, announced that its open-source agent application, DeepSeek Harness, reached 1.54 million daily active users by November 4. This growth occurred in less than three months following the app's debut on August 13. The platform has seen a sharp increase in engagement since the company released a desktop version of the software in late September.

According to reports from Free Malaysia Today, the user base was significantly smaller just a few weeks ago. At the end of September, the company recorded approximately 271,000 daily users connecting to its own servers. This rapid scaling suggests a growing appetite for tools that go beyond simple text generation to perform more complex actions.

DeepSeek is using this momentum to position itself as a leader in AI agent infrastructure. While the company previously gained attention for its low-cost R1 reasoning model, it is now focusing on software that allows AI to act autonomously. These agents are designed to complete tasks on a user's behalf with minimal human supervision.

The shift to autonomous agents

AI agents differ from standard chatbots because they can execute multi-step workflows. These systems can write and modify code, analyze various files, and conduct independent research. They are also capable of using other tools to complete objectives without needing a human to prompt every single step of the process.

DeepSeek is building Harness using what it calls an "everything is a plugin" architecture. A plugin is a piece of code that adds a specific feature to a software program without needing to change the core software itself. This approach allows the platform to expand its capabilities quickly by letting different tools be integrated into the agent's workflow.

The startup currently counts about 35,000 plugins within its ecosystem. Interestingly, the company reported that 60% of its users have built at least one of these plugins. While the complexity of these tools varies, the high participation rate shows a strong developer interest in building specialized tools on DeepSeek's infrastructure.

Escalating compute demands

The move toward autonomous agents has created a massive demand for computational power. DeepSeek reported that the Harness application consumes approximately 92 trillion tokens every day. A token is a unit of text that the AI processes to generate a response or perform a task.

This translates to about 60 million tokens being used daily for each active user. This figure provides some of the first concrete evidence that AI agents require much more processing power than traditional chatbots. Because agents run for multiple hours to read code, browse the web, and use various tools, they put a heavy strain on hardware resources.

This surge in demand highlights a major technical hurdle for the industry. As AI moves from simple conversation to autonomous action, the amount of compute required to sustain those actions grows exponentially. DeepSeek's data shows that the hunger for power is vastly higher when an AI is acting as an agent rather than just answering a question.

Competing on ecosystems

DeepSeek is using its open-source approach to challenge major players like OpenAI and Anthropic. Instead of just competing on model performance or price, the company is trying to win by building a broader ecosystem. They want to create a space where third-party applications and specialized tools can flourish around their technology.

The company plans to launch an official marketplace where developers can publish their plugins for others to find. This marketplace will be supported by a program that offers token grants, technical assistance, and promotional opportunities for creators. These moves are intended to build a loyal developer base that creates a competitive advantage for DeepSeek's infrastructure.

By focusing on integration, DeepSeek is betting that a diverse library of tools will make its platform more useful than proprietary systems. The goal is to move the industry beyond the chatbot phase and into a world where AI agents are integrated into every specialized workflow. This strategy aims to make DeepSeek a foundational layer for the next generation of AI software.