DeepSeek launching v4.1 flash cheaper and more capable than v4 pro

DeepSeek plans to launch V4.1 Flash around September 10, 2026, a model that outperforms V4 Pro across all key metrics while being cheaper and faster. During the transition period between V4.1 Flash's launch and V4.1 Pro's release, all Pro model requests will be automatically routed to V4.1 Flash and charged at Flash's lower pricing. The new pricing structure features tiered rates for off-peak and peak hours, with off-peak input cache hits at $0.003 and output at $0.6.