AI News

OpenRouter Releases a New Generation of Model Routers: Auto Router and Auto Code

OpenRouterUpdated Nov 7 00:37

AI summary

OpenRouter has launched a new generation of model routers: Auto Router (for general Q&A and agents) and Auto Code (for coding agents). According to the official announcement, Auto Code's benchmark performance is close to Opus 5.5, at 74% lower cost, with p90 latency of only 0.65s, and it can show the basis for routing decisions on a per-request basis. Auto Router processes 2.5T tokens generated by over 100k users each month; during the testing period, invalid tool calls were reduced by 4x, agent response truncation decreased by 50%, and latency was reduced by 3x compared with the previous generation of routers.

Why read this

AI score 62

The official release provides benchmark comparisons between Auto Code and Opus 5.5, cost and latency data, and a transparent view of routing decisions, making it easier to judge the router's actual trade-offs.

Topics

Tags

Body

It's easier for routers to achieve benchmarks than it is to perform well in practice. Auto Router sees over 2.5T tokens from over 100k people every month, so it must perform in the real world.

The new generation builds upon principles developed from 3 years of running Auto Router...

Source

OpenRouter