The global artificial intelligence race is getting more competitive, with Chinese AI startup DeepSeek introducing its latest V4 Pro AI model. The new model is positioned as another step in the company's efforts to compete with leading American AI developers, including OpenAI and Anthropic.

DeepSeek says the new V4-Pro-0813 model comes with improved AI agent capabilities, an area where major technology companies are investing heavily. The model has been made available through DeepSeek's API, mobile app and web platform.

DeepSeek V4 Pro Focuses on AI Agents

One of the biggest areas of focus for V4 Pro is its ability to work as an AI agent. Unlike a basic chatbot that primarily responds to individual prompts, agent-focused AI systems are designed to handle more complex, multi-step tasks and interact with tools and services.

DeepSeek says V4-Pro-0813 offers enhanced capabilities in this area, potentially making it more useful for developers and businesses building automated AI workflows.

The company has also revised API pricing for its V4 Pro and V4 Flash models, with different rates reportedly applying during peak and off-peak periods.

Interestingly, some independent evaluations have reportedly shown the lower-priced V4 Flash performing better than an earlier preview version of V4 Pro in certain tests. Benchmark results, however, can vary considerably depending on the tasks and testing methodology.

DeepSeek Faces Competition at Home and Abroad

DeepSeek became one of China's most closely watched AI companies after its R1 model attracted global attention in early 2025. The model sparked wider discussion about whether highly capable AI systems could be developed and operated at substantially lower costs.

Since then, competition within China's AI industry has intensified.

Companies including Moonshot AI, Zhipu AI, MiniMax, Alibaba and ByteDance have been developing and releasing their own models, creating an increasingly crowded domestic market.

DeepSeek therefore isn't competing only with American companies. It must also keep pace with a rapidly expanding group of Chinese AI developers.

Can DeepSeek Challenge ChatGPT and Claude?

DeepSeek's ambitions extend beyond China. With V4 Pro, the company is seeking to compete more directly with advanced AI products and models developed by OpenAI and Anthropic.

But competing in the global AI market involves more than achieving strong benchmark scores. Reliability, reasoning performance, coding capabilities, agent functionality, API costs, developer adoption and infrastructure all influence whether a model gains long-term traction.

OpenAI's ChatGPT and Anthropic's Claude already have significant global user and developer ecosystems, meaning DeepSeek will need to demonstrate consistent performance and practical advantages if it wants to win users away from established platforms.

DeepSeek Reportedly Preparing for Major Expansion

DeepSeek is also working to turn its rapid rise in popularity into a larger and more sustainable business.

According to the information cited in the report, the company has been exploring further fundraising while expanding its workforce.

DeepSeek reportedly plans to at least double its staff, with hiring focused particularly on AI agents and data-centre operations.

Another notable area is semiconductor development. The company is reportedly recruiting chip-design engineers as it explores ways to reduce dependence on external suppliers such as Nvidia and Huawei.

Developing greater control over computing infrastructure could become strategically important as access to advanced chips remains one of the biggest challenges facing AI companies.

Global AI Race Set to Become Even More Competitive

The arrival of DeepSeek V4 Pro highlights how quickly the AI landscape is changing. The competition is no longer limited to a small number of US technology companies, with Chinese developers increasingly introducing models designed to compete on performance, cost and specialised capabilities.

DeepSeek's next challenge will be proving that V4 Pro can deliver reliable real-world performance while building a strong developer ecosystem around it.

Whether it can seriously challenge established platforms such as ChatGPT and Claude will depend not just on model benchmarks, but also on pricing, infrastructure, usability, safety and how quickly developers and businesses adopt its technology.