Chinese AI lab DeepSeek released its V4 large language models to challenge industry leaders OpenAI and Google. The launch features two open-source versions: the high-performance DeepSeek-V4-Pro and the economical DeepSeek-V4-Flash. DeepSeek claims these models rival top closed-source systems in reasoning and coding at a fraction of the cost.
The V4-Flash model is priced at $0.14 per million input tokens. Output tokens for V4-Flash cost $0.28 per million.
The high-performance V4-Pro model costs $1.74 per million input tokens. Output tokens for V4-Pro are priced at $3.48 per million.
This aggressive pricing structure increases pressure on Microsoft Azure OpenAI and Google Gemini services. The strategy may commoditize the AI model market and accelerate adoption for cost-sensitive applications.