China’s DeepSeek Unveils GPT-4 Rival ‘DeepSeek-V2’ to Intensify AI Race with OpenAI

DeepSeek-V2 Launch:
Chinese AI company DeepSeek announced its upgraded large language model, DeepSeek-V2, aimed at competing directly with OpenAI’s GPT-4.

Model Specifications:
DeepSeek-V2 is built with 236 billion parameters, significantly enhancing its performance in comparison to previous models. It supports both English and Chinese language tasks.

Performance Claims:
The company claims DeepSeek-V2 achieves performance comparable to GPT-4 and outperforms GPT-3.5 in various benchmark tests, including reasoning, mathematics, and coding abilities.

Cost Efficiency:
DeepSeek-V2 is designed to offer high cost-efficiency in both training and inference stages, making it attractive for enterprise and developer use.

Open-Source Release:
The model is open-sourced to the global developer community, aligning with China’s broader AI development strategies and fostering global collaboration.

China-U.S. AI Rivalry:
This launch intensifies the AI race between China and the U.S., with Chinese firms accelerating efforts to rival tech giants like OpenAI and Anthropic.

Market Impact:
DeepSeek’s move signals China’s increasing capability in foundational AI technologies and large-scale LLMs, aiming to reduce reliance on Western AI tools.

Broader Implications:
The release is part of a growing trend of Chinese companies prioritising AI development, supported by significant domestic investments and government backing.