X @BSCN

DEEPSEEK V4 GOES LIVE, CLAIMS 3-6 MONTH GAP TO FRONTIERDeepSeek (@deepseek_ai) released preview versions of V4 Flash and V4 Pro earlier today. Both are mixture-of-experts models with 1 million token context windows.V4 Pro runs 1.6 trillion total parameters with 49 billion active, making it the largest open-weight model currently available. It beats Moonshot's Kimi K 2.6 (1.1T) and more than doubles last year's V3.2 (671B).The lab says V4 matches GPT-5.4 on coding benchmarks and outperforms some closed model ...