Top Story
9/10
High signal
DeepSeek Releases V4 Flagship Models: World-Class Reasoning at Rock-Bottom Prices
Chinese AI lab DeepSeek launched V4 Pro and V4 Flash, its latest open-weight frontier models with million-token context windows and benchmarks rivaling GPT-5.5, Claude Opus 4, and Gemini 3.1 Pro. Priced at just $0.14 per million input tokens for Flash, it undercuts Western competitors by up to 75%. The release intensifies the global AI arms race with high-performance, low-cost options.
Why it matters
A Chinese flagship model matching top US labs on performance while slashing costs could democratize advanced AI and challenge Big Tech dominance.