China’s DeepSeek releases long-awaited new AI model

Hong Kong Free Press reports on the significant release of DeepSeek-V4, a new artificial intelligence model from Chinese startup DeepSeek. This development marks another stride for China in the global AI race, particularly in the open-source domain.

“

features an ultra-long context of one million words

DeepSeek-V4 boasts an impressive one million-word context length, positioning it as a leader in agent capabilities, world knowledge, and reasoning. The availability of both a Pro and a more economical Flash version, along with optimizations for popular AI Agent products, underscores DeepSeek's strategy to offer powerful yet accessible AI solutions.

“

cost-effective

The article highlights the "DeepSeek shock" from last year, when the company's previous model challenged US dominance and spurred a re-evaluation of AI strategies globally. While DeepSeek's open-source approach contrasts with Western proprietary models and has led to wide adoption within China across various sectors, it has also raised questions about data privacy and censorship, particularly concerning sensitive topics.

“

achieves leadership in both domestic and open-source fields across agent capabilities, world knowledge, and reasoning performance

This release is particularly noteworthy from a Hong Kong perspective, as it showcases the rapid advancement of AI technology originating from mainland China. It underscores the growing competition and innovation in the AI landscape, where Chinese companies are increasingly playing a pivotal role, not just as users but as developers and innovators setting new benchmarks.

“

a more efficient and economical choice