CekuraBench

Azure Realtime

Every modelCompare

70.7%

Reliability

4th of 12

1.66s

Median reply

4th of 12

$0.080

Cost / min

4th 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.

Azure Realtime 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 Azure Realtime struggled

Azure Realtime falls furthest behind on background noise: 67% passed, against a median of 81%.

Below the medianPassedMedian
  • Background noise
    67%median 81%
  • Several requests in one call
    86%median 95%
Mistakes in saved data
  • Field left out45 · 55%
  • Step skipped14 · 17%
  • Wrong value12 · 15%

Fields it got wrong most: route_id, lead_id, route_status

Response time at P50, P90 and P95

Azure Realtime starts replying in 1.66s on a typical turn and 2.42s on the slowest turn in twenty, 4th 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.

Azure Realtime head to head

Every comparisonAll 12 models

How much audio Azure Realtime was tested on

246 live calls: 82 scenarios, each run three times. The median call ran 1:59.

Live calls
246
Scenarios
82
Booking calls
177
Medicare calls
69
Replies per call
7.0
Median call
1:59

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

Azure Realtime FAQ

  • Azure Realtime passed 58 of 82 scenarios on all three runs, 70.7%, 4th of 12 realtime models. Across all 246 calls, 84.1% passed every check.

  • Azure Realtime starts replying 1.66s after the caller finishes at the median, 2.22s at P90 and 2.42s at P95. That is 4th fastest of 12 at the median, measured on the call audio.

  • Azure Realtime falls furthest behind the field on background noise: 67% of calls passed, against a median of 81%. When it saved data wrong, the most common mistake was field left out, 55% of them.

  • Azure Realtime saved everything exactly right on 86.1% of the calls that save data, 4th of 12. On Booking calls, 94.2%. On Medicare calls, 65.2%.

  • $0.080 a minute of call at Microsoft's published rates, 4th cheapest of 9 priced models. Each call's metered tokens at Voice Live's Pro rates, read from Azure's price list on 7 Oct 2026. Excludes caller transcription, which Azure bills separately as speech input.

  • Cekura ran Azure Realtime 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 Azure Realtime. A call passes only when the agent said the right things and saved the right data.