API Provider Leaderboard
Grouped by API endpoint hostname and ranked by model match quality, leading GPT and Claude coverage, evidence volume, and recent testing activity.
Provider overall leaderboard
The overall score combines match quality, leading GPT / Claude coverage, model breadth, sample confidence, and recent activity.
How is the overall score calculated?
70%Current performance: at most the latest 15 diagnostics, so repeated testing cannot keep raising the score
30%Historical reputation: once there are more than 15 tests, older match quality still affects the total
40In-window match quality (with an 8-test neutral prior)
25In-window leading GPT / Claude coverage (12.5 each, tolerating isolated mismatches)
10In-window verified model breadth (capped at 6 models)
10Sample confidence (capped at the latest 15 diagnostics)
15Recent activity (30-day tests in the window plus last-test recency)