CekuraBench

Gemini 3.1 Flash Live Preview

Every modelCompare

62.2%

Reliability

7th of 12

2.88s

Median reply

12th of 12

$0.070

Cost / min

3rd of 9

Reliability vs. response time

Higher and further left is better

Best quadrant
Reliability, pass³ (%)
Median response time (s)

For the wider field, see our pick of the best voice AI APIs for real-time audio.

Gemini 3.1 Flash Live Preview against every other model

Scenarios that passed on all 3 runs, pass³. Higher is better.

#ModelReliability
  1. 01Grok Voice 3.081.7%
  2. 02GPT Realtime 2.179.3%
  3. 03GPT-Live 1 + GPT-6 Sol75.6%
  4. 04Azure Realtime70.7%
  5. 05Phonic v168.3%
  6. 06Grok Voice Think Fast 2.067.1%
  7. 07Gemini 3.1 Flash Live Preview62.2%
  8. 08Deepslate Opal57.3%
  9. 09Gemini 3.8 Live Extended Thinking52.4%
  10. 10Amazon Nova 2 Sonic50.0%
  11. 11GPT Realtime 2.1 Mini50.0%
  12. 12Gemini 3.8 Live41.5%
  • refCascade82.9%

Where Gemini 3.1 Flash Live Preview struggled

Gemini 3.1 Flash Live Preview falls furthest behind on Medicare compliance and routing: 32% passed, against a median of 40%.

Below the medianPassedMedian
  • Medicare compliance & routing
    32%median 40%
  • Silence, cough & packet loss
    83%median 88%
  • Interruptions & barge-in
    87%median 90%
Mistakes in saved data
  • Wrong value87 · 49%
  • Field left out48 · 27%
  • Extra call26 · 15%

Fields it got wrong most: lead_id, destination, route_id

Response time at P50, P90 and P95

Gemini 3.1 Flash Live Preview starts replying in 2.88s on a typical turn and 4.41s on the slowest turn in twenty, 12th fastest of 12 at the median.

ModelP50P90P95
  • Cascade
    2.17s2.51s2.62s
Seconds from the caller finishing to the agent replying. The bar is the typical turn, P50.

Gemini 3.1 Flash Live Preview head to head

Every comparisonAll 12 models

How much audio Gemini 3.1 Flash Live Preview was tested on

246 live calls: 82 scenarios, each run three times. The median call ran 1:37.

Live calls
246
Scenarios
82
Booking calls
177
Medicare calls
69
Replies per call
7.7
Median call
1:37

82 scenarios, the same for every model. See every test case

Gemini 3.1 Flash Live Preview FAQ

  • Gemini 3.1 Flash Live Preview passed 51 of 82 scenarios on all three runs, 62.2%, 7th of 12 realtime models. Across all 246 calls, 76.0% passed every check.

  • Gemini 3.1 Flash Live Preview starts replying 2.88s after the caller finishes at the median, 4.00s at P90 and 4.41s at P95. That is 12th fastest of 12 at the median, measured on the call audio.

  • Gemini 3.1 Flash Live Preview falls furthest behind the field on Medicare compliance and routing: 32% of calls passed, against a median of 40%. When it saved data wrong, the most common mistake was wrong value, 49% of them.

  • Gemini 3.1 Flash Live Preview saved everything exactly right on 79.2% of the calls that save data, 6th of 12. On Booking calls, 96.5%. On Medicare calls, 34.8%.

  • $0.070 a minute of call at Google's published rates, 3rd cheapest of 9 priced models. Averaged over the calls with a usage record.

  • Cekura ran Gemini 3.1 Flash Live Preview as the whole agent in the same open-source Pipecat pipeline as every other model, with the same prompt, tools and connection. A simulated caller worked through 82 scenarios on live calls, three times each: 246 calls for Gemini 3.1 Flash Live Preview. A call passes only when the agent said the right things and saved the right data.