DeepSeek's new AI model challenges GPT-5, optimized for Chinese chips
Chinese AI company DeepSeek just unveiled V3.1, its latest language model aimed at competing with OpenAI's GPT-5.
With 685 billion parameters and a smart "mixture-of-experts" setup (which basically means it only uses parts of the model needed for each task), V3.1 is designed to be powerful but efficient.
The cool part? It is optimized for Chinese-made chips, signaling resilience against US export controls.
Model's hybrid design earns praise for quick responses and reasoning
V3.1 isn't just about big numbers; its hybrid design delivers quick responses and better reasoning, earning praise from tech analyst Ben Dickson as a standout feature.
DeepSeek has priced it to go head-to-head with GPT-5, making waves in the industry and pushing into global markets beyond China.
Even OpenAI's CEO Sam Altman acknowledged that competition from Chinese open-source models like DeepSeek was a factor in their decision to release their own open-weight models—proof that this new player is shaking things up in the AI world.