Summary
On August 12, 2026 OpenRouter launched Auto, a new automatic model router informed by the model choices of millions of users across its marketplace. OpenRouter says Auto outperforms conventional task-based classifiers across a wide spectrum of tasks by picking a suitable model per request instead of relying on hand-tuned heuristics.
What changed
OpenRouter released Auto, a request-level router that selects a model per prompt using patterns learned from millions of aggregated user model decisions rather than static task classifiers.
Why it matters
OpenRouter's cross-vendor traffic gives it a routing signal few competitors can match, turning marketplace scale into a quality-and-cost optimization layer. As multi-model routing becomes the default abstraction, whoever routes best captures the developer relationship and the spend.
Evidence excerpt
OpenRouter's Auto router is "informed by the model decisions of millions of people and outperforms conventional task-based classifiers across a wide spectrum of tasks."