Updated
Updated · Engadget · Aug 14
DeepSeek Quadruples V4 Pro API Price to $3.96 per Million Tokens
Updated
Updated · Engadget · Aug 14

DeepSeek Quadruples V4 Pro API Price to $3.96 per Million Tokens

3 articles · Updated · Engadget · Aug 14

Summary

  • $3.96 per 1 million output tokens is what DeepSeek will charge for V4 Pro at peak hours from Aug. 16, up from $0.87; off-peak pricing will be $1.98.
  • DeepSeek said the new peak and off-peak structure is meant to allocate resources more reasonably as it rolls out its latest model, after warning customers to prepare for steep increases.
  • V4 Flash will also rise to $1.32 per 1 million output tokens at peak hours from $0.28, with off-peak usage priced at $0.66.
  • The move reverses DeepSeek's earlier decision to keep promotional API discounts permanent, though its V4 models still undercut Moonshot's Kimi K3 at $15 and OpenAI's GPT-5.6 Sol at $30; only GPT-5.6 Luna is cheaper at $1.20.

Insights

As DeepSeek quadruples its peak rates, is the golden era of subsidized frontier AI officially dead?
Could this massive price hike be the catalyst that finally pushes enterprises to build their own local AI infrastructure?
With ultra-cheap AI vanishing, will developers be forced to abandon complex reasoning models to survive the coming cost crisis?