DeepSeek's latest model fails to close AI gap with top US models

China startup Deepseek's V4 fails to close the AI capability gap with frontier US models, despite offering high performance at a lower cost. "V4 performance falls marginally short of GPT-5.4 and Gemini-3.1-Pro, suggesting a developmental trajectory that trails state-of-the-art frontier models by…3 to 6 months," Deepseek said. V4 series supports text only, unlike many of its closed-source peers.