Grok Voice 3.0 vs GPT Realtime 2.1 Mini

Grok Voice 3.0 is more reliable, by 28.6 points. Grok Voice 3.0 handled 30 of 70 real callers well on all three runs, and GPT Realtime 2.1 Mini 10. A caller only counts when every run of the call scored 80 or more.

For a voice agent, Grok Voice 3.0, since it handles more real callers well on every run. Grok Voice 3.0 is ahead on ELO rating, hearing score, turn taking score, speaking score, task handling score, reliability, calls handled well, calls fully resolved, calls where the agent went quiet and median latency, and GPT Realtime 2.1 Mini on cost.

Grok Voice 3.0, by 21.0 points higher. Grok Voice 3.0 fully resolved 34.8% of calls and GPT Realtime 2.1 Mini 13.8%. A call counts only when every step the caller needed was done.

Yes. Grok Voice 3.0 starts replying 0.80s sooner at the median. Grok Voice 3.0 takes 1.60s at the median and 1.94s at p90; GPT Realtime 2.1 Mini takes 2.40s and 2.88s. Both are measured on the call audio, from the caller finishing to the agent starting to reply.

Grok Voice 3.0, by 1.9 points. The agent went quiet for 10 seconds or more on 7.1% of Grok Voice 3.0 calls and 9.0% of GPT Realtime 2.1 Mini calls.

GPT Realtime 2.1 Mini is $0.060 cheaper a minute. Grok Voice 3.0 costs $0.080 a minute of call and GPT Realtime 2.1 Mini $0.020, at each vendor's published rates.

Cekura ran both as the whole agent with the same prompt, tools and phone connection. Both took the same 70 recorded real human callers on live calls, three times each. Each call is graded on 25 skills across Hearing, Turn Taking, Task Handling and Speaking, and passes at a score of 80 out of 100.