Key Info

OpenRouter introduced a new routing algorithm that gives more traffic to providers delivering strong inference performance. Over the past month, GLM 5.3 Flash users saw TTFT fall 40%, throughput rise 10%, and cost per 1M tokens drop 17%.

Highlights

  • New routing algorithm rewards providers with better inference by allocating more traffic to them
  • GLM 5.3 Flash TTFT down 40% over the past month
  • Throughput up 10% for the same model
  • Price per 1M tokens down 17%
  • OpenRouter frames the result as faster, more reliable, and cheaper inference