model · DeepSeek

DeepSeek V4

DeepSeek V4-Flash is a frontier-class model at a fraction of the usual price — roughly $0.14 per 1M input tokens, around 35x cheaper than the big US flagships — with a 1M-token context and a thinking mode. DeepSeek publishes its model weights (reported under the permissive MIT licence), though the full models are cluster-scale, so most people use the cheap hosted API. It is text-only. When cost or open weights matter most, DeepSeek is the value pick.

  • paid
  • self-host
  • Commercial

What it can help with

  • frontier model
  • large language model
  • high context window
  • open weights
  • low cost api
  • reasoning mode