A translated meme circulating on Reddit indicates that a competitor had to reduce their pricing by 80% due to the competitive pressure from DeepSeek v4 flash. This open weights model features 284 billion total parameters with 13 billion active parameters, delivering superior price-performance metrics.

  • The source is a translated image post shared on Reddit regarding market reactions to DeepSeek's new model.
  • DeepSeek v4 flash utilizes a Mixture of Experts architecture with 284B total and 13B active parameters.
  • Competitors reportedly cut prices by 80% in response to this specific price-performance benchmark.