news•23 Dec 2025China's DeepSeek V3: Open-Source Model Matches GPT-4 at Fraction of CostChinese AI startup DeepSeek releases V3, an open-source model that rivals GPT-4 performance while costing 95% less to train.
news•23 Dec 2025Meta Llama 4: Open-Source Model Reaches Frontier PerformanceMeta's Llama 4 closes the gap with GPT-4 and Claude while remaining fully open-source, reshaping the AI landscape for developers and researchers.