Key Info

Particle-1, a coding model post-trained on DeepSeek V4.1 Flash by Particle, is now live on OpenRouter and is positioned around token efficiency rather than raw benchmark gains.

Highlights

  • Particle reports 28–52% fewer output tokens than DeepSeek V4.1 Flash across four coding benchmarks
  • Average token reduction is roughly 45%, with accuracy said to be close to the base model
  • Lower token use means agent tasks finish sooner and cost less
  • Pricing is $0.25/M input tokens and $1/M output tokens
  • Context window is 1.05M tokens