ALBAWABA – DeepSeek, the innovative Chinese artificial intelligence company dedicated to advancing AI and natural language processing, announced the release of DeepSeek V3, a new AI model that sets it apart from competitors in the market.
DeepSeek V3
DeepSeek V3 was released with significant features that outperform ChatGPT and Llama.
Developers can now download and modify the new DeepSeek V3 model to perform multiple tasks, including coding, translation, and writing articles and emails, according to TechCrunch and Al Arabiya.
The company stated that the new model outperformed many well-known artificial intelligence (AI) models on the market, which are typically only accessible via an API.
DeepSeek V3 was released with significant features that outperform ChatGPT and Llama. (DeepSeek)
Notably, DeepSeek surpassed other models, including Meta's Llama 3.1 (405B), OpenAI's GPT-4, and Alibaba's Qwen 2.5 (72B), in programming competitions hosted on the Codeforces platform.
Additionally, DeepSeek V3 excelled in the Aider Polyglot competition, a test designed to measure whether a model can successfully write new code that integrates with existing code.
According to DeepSeek, DeepSeek V3 AI model was tested multiple times on a 14.8 trillion data set, where one million tokens are equivalent to about 750,000 words.
