Compare speech-to-speech (S2S) benchmarks
- Reliability90.2%82.9%
- Success rate95.1%93.1%
- Data accuracy97.5%94.1%
- Went quiet0.0%0.8%
- Response time1.63s2.17s
- Cost / min$0.0600n/a
- Measures won5 of 60 of 6
Any two of the 14 models, or any one against the cascade, on the same voice agent and the same calls.
Last updated Methodology by Dileep Chagam
By provider
The best model from each provider, metric by metric. For the wider field, see our pick of the best voice AI APIs for real-time audio.
Every comparison
Phonic v1
Grok Voice 3.0
GPT Realtime 2.1
GPT-Live 1 + GPT-6 Sol
Azure Realtime
Prism
Grok Voice Think Fast 2.0
Gemini 3.1 Flash Live Preview
Deepslate Opal
Higgs Realtime
Gemini 3.8 Live Extended Thinking
Amazon Nova 2 Sonic
GPT Realtime 2.1 Mini
Gemini 3.8 Live
CascadeNot ranked
Speech-to-speech model comparison FAQ
How do I compare speech-to-speech models?
Choose any two models to compare repeatable reliability, task success, saved-data accuracy, going quiet, response time, interruption handling, and cost from the same live-call benchmark.
Which metrics matter most for a realtime voice model?
Start with repeatable task reliability and response time, then check data accuracy, going quiet, interruption handling, and cost. A fast model is not useful if it cannot complete the same scenario reliably.
Can I compare GPT Realtime with Gemini Live?
Yes. Select the tested GPT Realtime and Gemini Live versions to compare their reliability and latency on the same voice agent, prompt, tools, and caller scenarios.
What is the difference between a native speech-to-speech model and a cascade?
A native model handles audio directly. A cascade connects speech-to-text, an LLM, and text-to-speech. The comparison pages test both as the same complete agent so their practical trade-offs are visible.
Test your voice agent the same way.
Cekura runs these simulated calls on your agent and scores every one.