DeepSeek V4 Pricing Undercuts Every Frontier Model: Flash at $0.14/M Input, Pro at $0.145/M Input with $0.20/M Cache Hits

DeepSeek API Docs·medium signal

DeepSeek V4 Flash costs $0.14 per million input tokens and $0.28 per million output tokens, undercutting GPT-5.4 Nano, Gemini 3.1 Flash, and Claude Haiku 4.5. V4 Pro costs $0.145/$3.48 per million tokens, cheaper than GPT-5.5, Claude Opus 4.7, and Gemini 3.1 Pro. Cache-hit pricing drops Flash input to $0.20/M tokens. The API supports OpenAI ChatCompletions and Anthropic interfaces — switching requires only changing the model name. For production agent pipelines, V4 Flash with caching may be the cheapest frontier-capable inference available.