Race the newest text, speech and voice models — on live APIs.
Measure time-to-first-token, time-to-first-audio, throughput, real-time factor, word error rate and cost side by side, with p50/p90 over repeated runs.
TTFT / TTFA
ms to first token or audio
Throughput
tok/s and ×RT
Accuracy
WER vs reference
Spend
$ per call from list price
Internal tool · Synthetic data only — never paste PHI.
CallSphere War Room
Model benchmarking
Sign in
Use your CallSphere platform account.
Access is limited to allowlisted CallSphere staff with two-factor authentication. Forgot your password? Reset it at app.callsphere.ai.