Phonic v1 is ahead on reliability, success rate, agent responses, data accuracy, stalled calls and response time, and GPT Realtime 2.1 Mini on interruption and cost.
Scenarios that passed on all three runs.
Phonic v1, 17.1 points higher
Calls that passed every check.
Phonic v1, 12.6 points higher
Calls where the agent's replies reached the expected outcome.
Phonic v1, 4.5 points higher
Calls where everything the agent saved matched exactly.
Phonic v1, 9.7 points higher
Calls where the agent went silent for 10 seconds or more.
Phonic v1, 2.5 points lower
How well it handled being talked over, 0 to 5.
Realtime 2.1 Mini, 0.11 higher
Median time to start replying after the caller finishes.
Phonic v1, 0.49s sooner
One minute of call at the vendor's published rates.
Realtime 2.1 Mini, $0.120 cheaper
The small figure is the place among ranked models. The cascade is not ranked.
Phonic v1 1.59s, Realtime 2.1 Mini 2.08s at the median.
GPT Realtime 2.1 Mini passed 41 of 82 scenarios on all three runs, and Phonic v1 passed 55. Phonic v1 is ahead, by 17.1 points.
GPT Realtime 2.1 Mini saved everything exactly right on 64.5% of the calls that save data, and Phonic v1 on 74.3%. On Appointments, 88.8% against 98.2%. On Medicare, 0.0% against 12.1%. A call counts only when nothing is missing, wrong, extra or saved twice.
Phonic v1 starts replying 0.49s sooner at the median. GPT Realtime 2.1 Mini takes 2.08s at the median and 2.75s at p90; Phonic v1 takes 1.59s and 1.86s. Both are measured on the call audio, from the caller finishing to the agent starting to reply.
GPT Realtime 2.1 Mini costs $0.020 a minute of call and Phonic v1 $0.140, at each vendor's published rates. GPT Realtime 2.1 Mini is $0.120 cheaper a minute.
Both ran as the whole agent in the same open-source Pipecat pipeline, with the same prompt, tools and phone line. A simulated caller worked through all 82 scenarios on live calls, three times each, 2,214 calls across the benchmark. A call passes only when the agent said the right things and saved the right data.
