- Reliability
- 70.7%
- Response time
- 1.66s
- Cost / min
- $0.0798
Azure Realtime is ahead on reliability, success rate, agent responses, data accuracy and response time, and Amazon Nova 2 Sonic on calls where the agent went quiet and interruption.
Last updated Methodology by Dileep Chagam
Our guide covers testing OpenAI Realtime API voice agents.
Scenarios the agent passed on every one of its 3 runs.
Each model on its own: Azure Realtime and Amazon Nova 2 Sonic
Azure Realtime is more reliable, by 20.7 points. Azure Realtime passed 58 of 82 scenarios on all three runs, and Amazon Nova 2 Sonic passed 41. A scenario only counts when every run of it passed.
Azure Realtime is more reliable, by 20.7 points. Azure Realtime passed 58 of 82 scenarios on all three runs, and Amazon Nova 2 Sonic passed 41. A scenario only counts when every run of it passed.
For a voice agent, Azure Realtime, since it gets more scenarios right on every run. Azure Realtime is ahead on reliability, success rate, agent responses, data accuracy and response time, and Amazon Nova 2 Sonic on calls where the agent went quiet and interruption.
Azure Realtime, by 26.2 points. Azure Realtime saved everything exactly right on 86.1% of the calls that save data, and Amazon Nova 2 Sonic on 59.9%. On Booking calls, 94.2% against 82.5%. On Medicare calls, 65.2% against 1.5%. A call counts only when nothing is missing, wrong, extra or saved twice.
Yes. Azure Realtime starts replying 0.53s sooner at the median. Azure Realtime takes 1.66s at the median and 2.22s at p90; Amazon Nova 2 Sonic takes 2.18s and 2.76s. Both are measured on the call audio, from the caller finishing to the agent starting to reply.
Amazon Nova 2 Sonic, by 0.03 on a 0 to 5 scale. Azure Realtime scores 4.96 and Amazon Nova 2 Sonic 4.99 for how well the agent handled being talked over.
Amazon Nova 2 Sonic, by 2.8 points. The agent went quiet for 10 seconds or more on 4.1% of Azure Realtime calls and 1.2% of Amazon Nova 2 Sonic calls.
There is no verified per-minute rate for Amazon Nova 2 Sonic, so cost cannot be compared yet.
Cekura ran both as the whole agent in the same open-source Pipecat pipeline, with the same prompt, tools and connection. A simulated caller worked through all 82 scenarios on live calls, three times each, 3,936 calls across the benchmark. A call passes only when the agent said the right things and saved the right data.
Cekura runs these simulated calls on your agent and scores every one.