December 26, 2024:
DeepSeek-V3 Outshines Llama and Qwen on Debut - Chinese AI startup DeepSeek has introduced DeepSeek-V3, an open-source model with 671B parameters, using a mixture-of-experts architecture for efficient task handling. It is benchmarked as the strongest open-source AI, surpassing models like Meta's Llama-3.1 and Qwen, especially in Chinese and math-related benchmarks.
Remarkably, it outperformed the closed-source GPT-4o in most areas, except English-centric tasks. Available on Hugging Face and GitHub, DeepSeek aims to bridge the gap between open-source and closed-source AI by offering competitive performance and cost-efficient training.
