DeepSeek has officially launched its V4 Pro model and simultaneously introduced a new pricing structure for its API. Starting on August 16, the company implemented dynamic peak and off-peak rates to better manage server load and infrastructure costs. During peak hours, the cost for output tokens of V4 Pro rises to $3.96 per million, while off-peak rates drop to $1.98. A similar adjustment applies to the lighter V4 Flash model. Despite the increases, DeepSeek’s prices remain competitive compared to some Western rivals. The upgrade also includes performance improvements, with V4 Pro showing better results in tasks involving coding, scientific reasoning, and agent-based workflows compared to its predecessor.
About the Author
