- Reliability
- 81.7%
- Response time
- 1.34s
- Cost / min
- $0.0800
Grok Voice 3.0 is ahead on reliability, success rate, agent responses, data accuracy, interruption and response time, and PolyAI Dialog-RSN-1 on cost. The two tie on calls where the agent went quiet.
Last updated Methodology by Dileep Chagam
Scenarios the agent passed on every one of its 3 runs.
Grok Voice 3.0 is more reliable, by 17.1 points. Grok Voice 3.0 passed 67 of 82 scenarios on all three runs, and PolyAI Dialog-RSN-1 passed 53. A scenario only counts when every run of it passed.
Grok Voice 3.0 is more reliable, by 17.1 points. Grok Voice 3.0 passed 67 of 82 scenarios on all three runs, and PolyAI Dialog-RSN-1 passed 53. A scenario only counts when every run of it passed.
For a voice agent, Grok Voice 3.0, since it gets more scenarios right on every run. Grok Voice 3.0 is ahead on reliability, success rate, agent responses, data accuracy, interruption and response time, and PolyAI Dialog-RSN-1 on cost. The two tie on calls where the agent went quiet.
Grok Voice 3.0, by 14.3 points. Grok Voice 3.0 saved everything exactly right on 87.3% of the calls that save data, and PolyAI Dialog-RSN-1 on 73.0%. On Booking calls, 98.2% against 97.7%. On Medicare calls, 59.1% against 9.1%. A call counts only when nothing is missing, wrong, extra or saved twice.
Yes. Grok Voice 3.0 starts replying 0.02s sooner at the median. Grok Voice 3.0 takes 1.34s at the median and 1.62s at p90; PolyAI Dialog-RSN-1 takes 1.36s and 1.87s. Both are measured on the call audio, from the caller finishing to the agent starting to reply.
Grok Voice 3.0, by 0.13 on a 0 to 5 scale. Grok Voice 3.0 scores 4.98 and PolyAI Dialog-RSN-1 4.85 for how well the agent handled being talked over.
They go quiet on the same share of calls. The agent went quiet for 10 seconds or more on 0.0% of Grok Voice 3.0 calls and 0.0% of PolyAI Dialog-RSN-1 calls.
PolyAI Dialog-RSN-1 is $0.0394 cheaper a minute. Grok Voice 3.0 costs $0.0800 a minute of call and PolyAI Dialog-RSN-1 $0.0406, at each vendor's published rates.
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.