Open Methodology

About AIBench.cc

AIBench.cc is an open and neutral LLM API health benchmark. It measures P50/P95/P99 latency, TTFT, prompt cache, model purity, rate limits and real token cost.

Methodology

AIBench runs consistent probes for response time, usage fields, cache fields, rate limit headers and model identity signals. Model purity is graded with multiple evidence points.

Neutrality

Leaderboard ranking is based on anonymized aggregate data. Sponsors do not buy placement or influence scoring.

Privacy

API keys are used only for the current proxy request and are not written to reports, logs or databases. Temporary test keys are recommended.