← All leaderboards
Sourced answers
What this board tests: a question in, a synthesized answer with cited sources out. Every provider below ran the identical corpus on 2026-06-21. The composite blends measured quality, latency p50, cost per successful call and error rate. Method and corpora: how we measure.
| Rank | Provider | Composite | Correctness | Cost | Latency p50 | Field note |
|---|---|---|---|---|---|---|
| 1 | 7.9/10 | 1.00 | $0.005 / call | 1.5 s | Citation faithfulness ranks highest. | |
| 2 | 7.79/10 | 1.00 | $0.02 / call | 4.3 s | ||
| 3 | 7.58/10 | 1.00 | $0.01 / call | 16.3 s | ||
| 4 | 7.13/10 | 1.00 | $0.008 / call | 1.4 s | ||
| 5 | 7.02/10 | 1.00 | $0.06 / call | 2.4 s | Accurate answers, though per-query cost far exceeds peers. | |
| 6 | 6.91/10 | 1.00 | $0.005 / call | 2.5 s | ||
| 7 | 6.61/10 | 1.00 | $0.02 / call | 2.6 s |
Low scores are published alongside high ones — a ranking you can only trust if nothing gets hidden. The gateway's fee sits on top-ups, not per call, so the order above earns us nothing either way.