Rime Prism vs Cascade

Cascade is more reliable, by 14.6 points. Rime Prism passed 56 of 82 scenarios on all three runs, and Cascade passed 68. A scenario only counts when every run of it passed.

On reliability, no: the cascade passes more scenarios on every run. Cascade is ahead on reliability, success rate, agent responses, data accuracy and interruption, and Rime Prism on calls where the agent went quiet and response time. Both ran as the same agent, with the same prompt and tools.

Cascade, by 9.6 points. Rime Prism saved everything exactly right on 84.5% of the calls that save data, and Cascade on 94.1%. On Booking calls, 92.4% against 98.8%. On Medicare calls, 64.2% against 81.8%. A call counts only when nothing is missing, wrong, extra or saved twice.

Yes. Rime Prism starts replying 0.83s sooner at the median. Rime Prism takes 1.34s at the median and 1.57s at p90; Cascade takes 2.17s and 2.51s. Both are measured on the call audio, from the caller finishing to the agent starting to reply.

Cascade, by 0.05 on a 0 to 5 scale. Rime Prism scores 4.91 and Cascade 4.96 for how well the agent handled being talked over.

Rime Prism, by 0.8 points. The agent went quiet for 10 seconds or more on 0.0% of Rime Prism calls and 0.8% of Cascade calls.

There is no verified per-minute rate for Rime Prism and Cascade, 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, 3,690 calls across the benchmark. A call passes only when the agent said the right things and saved the right data. The cascade runs the same agent for reference and is never ranked.