Key Info
OpenRouter's next-generation Auto Router is cache-aware: because switching models forces the new model to rebuild the input token cache, the router prefers changing reasoning effort when that is sufficient and only swaps models when the gain outweighs the lost cache.
Highlights
- Switching models incurs extra cost while the new model rebuilds the input token cache.
- Auto Router first tries adjusting reasoning effort, switching models only when the benefit justifies the cache loss.
- When it does switch, it weighs live prices, benchmark performance, user guardrails and more to choose both model and reasoning effort.
- Part of OpenRouter's next-gen routing lineup alongside Auto Code for coding agents.