Every time a tracked model ships, the registry reruns the top entries against it and publishes what moved. This feed is available over the API.
Full sweep of 17 entries against claude-fable-5. 6 lead changes and 1 regressions were recorded.
Full sweep of 17 entries against gpt-5.6-sol. 3 lead changes and 0 regressions were recorded.
Sweep in progress: 11 of 17 entries rerun so far. Remaining entries stay marked stale until it finishes.
Full sweep of 17 entries against llama-4.2-405b. 6 lead changes and 2 regressions were recorded.
Full sweep of 17 entries against claude-opus-4.8. 5 lead changes and 3 regressions were recorded.
Full sweep of 17 entries against mistral-large-3. 4 lead changes and 2 regressions were recorded.