DeepSeek gets an update: More memory and competition with OpenAI

The Chinese company DeepSeek has released an update to its neural network — DeepSeek V3.1, which significantly enhances its capabilities. According to 3DNews, citing DeepSeek’s WeChat page, the new version of the model features an expanded context window, enabling it to process larger queries and maintain long conversations while preserving context. This strengthens DeepSeek’s position in competition with Western leaders such as OpenAI and highlights the success of Chinese developers in the field of artificial intelligence.

What’s New in DeepSeek V3.1?

The main improvement in DeepSeek V3.1 is the expansion of the context window, which determines how much information the model can take into account when processing requests. This allows the model to:

• Process large texts, such as lengthy documents or code.
• Support extended conversations without losing the meaning of previous messages.
• Improve response quality in tasks requiring deep contextual analysis, such as programming or data analytics.

Technical details of the update have not yet been disclosed, but the emphasis on increased memory underscores DeepSeek’s ambition to build versatile and powerful AI solutions.

Competition with Western Counterparts

DeepSeek has already established itself as a serious player in the AI market. Its DeepSeek R1 model, introduced in early 2025, outperformed several Western counterparts, including OpenAI’s solutions, across a number of benchmarks.

Key advantages of DeepSeek include:

• High speed: The model delivers fast processing of queries, making it attractive for developers and businesses.
• Affordability: DeepSeek offers competitive solutions at lower costs compared to American companies.
• Efficiency: The company achieves strong results with fewer resources, demonstrating the innovative approach of Chinese developers.

These factors have helped DeepSeek gain global popularity, especially in areas requiring the processing of large volumes of data, such as scientific research and software development.

Delay of DeepSeek R2: What’s Holding It Back?

Following the success of DeepSeek R1, Chinese media have been actively discussing expectations surrounding the next model — DeepSeek R2. However, its release has been delayed, with experts pointing to several reasons:

• Leadership perfectionism: DeepSeek CEO Liang Wenfeng is determined to perfect the model, which slows down development.
• Hardware shortages: Limited access to Nvidia graphics accelerators, widely used for AI training, creates a bottleneck.
• Local alternatives: Huawei chips used by DeepSeek still lag behind Western counterparts in terms of efficiency.

Despite these challenges, DeepSeek continues to strengthen its position, and the V3.1 update shows that the company is not slowing down.

Significance for the AI Market

The DeepSeek V3.1 update underscores China’s growing role in the global AI race. Expanding the context window makes the model more versatile and suitable for complex tasks such as text analysis, code generation, and process automation. Competition with OpenAI and other Western giants demonstrates that Chinese companies can create high-performance solutions with fewer resources — a factor that could shift the balance of power in the market.

In addition, DeepSeek’s success is stimulating the development of local AI ecosystems in China, including chip production and software development. This could reduce dependence on Western technologies in the future.

In Brief…

DeepSeek V3.1, with its expanded context window, strengthens the company’s position against market leaders like OpenAI. The update, announced in August 2025, makes the model more powerful for handling big data and long-form conversations. The success of DeepSeek R1 and the anticipation surrounding R2 highlight the ambitions of Chinese developers, despite challenges tied to hardware shortages and high quality standards. This update confirms that DeepSeek remains one of the leaders in building affordable and efficient AI solutions capable of reshaping the industry.