Chinese AI firm DeepSeek released its latest model, DeepSeek V3, on Wednesday.

The model is available under a permissive license that allows it to be modified and used for various applications, including commercial projects.

DeepSeek V3 outperformed other AI models like Meta’s Llama 3.1, OpenAI’s GPT-4, and Alibaba’s Qwen 2.5 in tasks such as coding, translation, and content generation.

With 671 billion parameters and trained on 14.8 trillion tokens, DeepSeek V3 was developed in two months for US$5.5 million.

It avoids sensitive topics per China’s regulations promoting “core socialist values.”