DeepSeek V4 Flash is the lower-latency V4 model for high-volume chat, coding, tool-use, and long-context workloads, supporting both thinking and non-thinking modes with a 1M token context window.
No benchmark data available.
