DeepSeek’s massive 75% price cut on its flagship AI model could signal a major shift in the global AI race. Behind the scenes, Huawei’s growing AI chip ecosystem may be starting to reshape how cheaply powerful AI can actually run.
The Chinese AI lab came to prominence in early 2025 after launching a large language model that trained on a fraction of the compute power and at a fraction of the cost of the big U.S. models like those from OpenAI and Anthropic.
DeepSeek V4-Pro scores 3,206 on Codeforces, ahead of GPT-5.4 and Gemini, while costing $3.48 per million tokens versus Claude’s $25, making it one of the most price-competitive frontier-class AI releases in 2026.
DeepSeek says both models are more efficient and performant than DeepSeek V3.2 due to architectural improvements, and have almost “closed the gap” with current leading models, both open and closed, on reasoning benchmarks.
Anthropic accuses DeepSeek, Moonshot, and MiniMax of using 24,000 fake accounts to distill Claude’s AI capabilities, as U.S. officials debate export controls aimed at slowing China’s AI progress.
DeepSeek is hiring for DeepSeek AI search, a multilingual, multimodal engine that could challenge Google’s search habit. The listings also point to persistent AI agents, signaling a broader push beyond chatbots.
The post DeepSeek AI search is the clear…
The company said that the model was trained on 15 trillion mixed visual and text tokens.
A Google study finds advanced AI models mimic collective human intelligence by using internal debates and diverse reasoning paths, reshaping how future AI systems may be designed.
The post Google Research suggests AI models like DeepSeek exhibit collec…