CekuraBench
Book a demo

Gemini 3.1 Flash Live Preview vs GPT Realtime 2.1

All comparisonsLeaderboard
Compare
Gemini 3.1 Flash Live PreviewGoogle
Wins 7 of 11
Reliability (Pass³)
34.3%
Latency P50
3.06s
Cost per Minute (USD)
$0.070
GPT Realtime 2.1OpenAI
Wins 4 of 11
Reliability (Pass³)
32.9%
Latency P50
2.16s
Cost per Minute (USD)
$0.085

Gemini 3.1 Flash Live Preview is ahead on ELO rating, turn taking score, speaking score, reliability, calls handled well, calls where the agent went quiet and cost, and GPT Realtime 2.1 on hearing score, task handling score, calls fully resolved and median latency.

Last updated Methodology by Dileep Chagam

Who is better at what

Gemini 3.1 Flash Live PreviewGPT Realtime 2.1
  • ELO Rating10621054
  • Hearing6872
  • Turn Taking7975
  • Speaking8079
  • Task Handling7778
  • Reliability (Pass³)34.3%32.9%
  • Pass Rate (score ≥ 80)58.6%56.7%
  • Task Completion Rate24.8%34.3%
  • Dead-Air Rate (≥ 10 s)7.6%11.9%
  • Latency P503.06s2.16s
  • Cost per Minute (USD)$0.070$0.085

Results by call

Right every time, on booking and Healthcare calls

Banking callsHealthcare calls
  1. 29%
    40%
    Gemini 3.1 Flash Live Preview
  2. 26%
    40%
    GPT Realtime 2.1

Related Comparisons

Each model on its own: Gemini 3.1 Flash Live Preview and GPT Realtime 2.1

Gemini 3.1 Flash Live Preview vs GPT Realtime 2.1 FAQ

  • Gemini 3.1 Flash Live Preview is more reliable, by 1.4 points. Gemini 3.1 Flash Live Preview handled 24 of 70 real callers well on all three runs, and GPT Realtime 2.1 23. A caller only counts when every run of the call scored 80 or more.

  • For a voice agent, Gemini 3.1 Flash Live Preview, since it handles more real callers well on every run. Gemini 3.1 Flash Live Preview is ahead on ELO rating, turn taking score, speaking score, reliability, calls handled well, calls where the agent went quiet and cost, and GPT Realtime 2.1 on hearing score, task handling score, calls fully resolved and median latency.

  • GPT Realtime 2.1, by 9.5 points higher. Gemini 3.1 Flash Live Preview fully resolved 24.8% of calls and GPT Realtime 2.1 34.3%. A call counts only when every step the caller needed was done.

  • No. GPT Realtime 2.1 starts replying 0.90s sooner at the median. Gemini 3.1 Flash Live Preview takes 3.06s at the median and 3.86s at p90; GPT Realtime 2.1 takes 2.16s and 2.70s. Both are measured on the call audio, from the caller finishing to the agent starting to reply.

  • Gemini 3.1 Flash Live Preview, by 4.3 points. The agent went quiet for 10 seconds or more on 7.6% of Gemini 3.1 Flash Live Preview calls and 11.9% of GPT Realtime 2.1 calls.

  • Gemini 3.1 Flash Live Preview is $0.015 cheaper a minute. Gemini 3.1 Flash Live Preview costs $0.070 a minute of call and GPT Realtime 2.1 $0.085, 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.

Test your voice agent the same way.

Cekura runs these simulated calls on your agent and scores every one.

Book a demoStart free