China's DeepSeek Releases New AI Models with 'World-Class' Reasoning
Chinese AI company DeepSeek released new artificial intelligence models called V4 Pro and Flash on Friday. The company claims these models have "world-class" reasoning abilities and can handle one million tokens of context - meaning they can remember much longer conversations.
Chinese AI startup DeepSeek announced its latest artificial intelligence models on Friday, about a year after its previous version became the top-rated free app on Apple's App Store in the United States.
The new V4 models come in two versions: "Pro" for maximum performance and "Flash" for faster responses at lower cost. Both can process up to one million tokens of context, which means they can remember much longer conversations and documents than previous versions.
DeepSeek says the V4 has the best coding abilities among open-source AI models. The models can also perform complex tasks automatically without human guidance - what experts call "agentic" capabilities.
The Flash version offers similar reasoning power to the Pro model but responds faster and costs less to run. DeepSeek promotes this as "cost-effective" AI that competes with expensive services from American companies.
DeepSeek made waves last year when it showed that Chinese companies could build AI models that rival those from Silicon Valley giants like OpenAI and Google, often at much lower cost.
Better AI models could mean smarter chatbots and coding assistants that regular people use every day. DeepSeek's models are free and open-source, potentially giving everyone access to advanced AI without paying high subscription fees like ChatGPT Plus.
Watch for user testing and comparisons with ChatGPT and other AI models to see if DeepSeek's claims hold up.
Was this article helpful?
0 people found this helpful