Key Info

Command Code has made Hy4 Flash available on all plans, a 1M-token context model with 224B total parameters and 12B active, priced at $0.070 input / $0.281 output / $0.003 cache per million tokens.

Highlights

  • Pricing is claimed to be roughly 2x cheaper than DeepSeek V4.1 Flash and GLM Flash, and about 10x cheaper on cache versus g1m-5.3 Flash
  • 1M token context window with a 224B total / 12B active parameter split
  • Author says accuracy benchmarking is still in progress but expects it to be a solid everyday work model
  • Author also jokes about perceived throughput, noting it feels like 2K TPS even when it isn't