CekuraBench
Book a demo

Gemini 3.8 Live

Every modelCompare

968

ELO

9th of 12

27.1%

Pass³

8th of 12

2.01s

Latency

5th of 12

ELO Rating vs. Latency P50

Higher wins more head-to-head calls (1000 = average); further left replies faster

Best quadrant
ELO Rating
Median latency (s)

Gemini 3.8 Live Against Every Other Model

Each model took the same recorded human callers on live calls. Pick a measure to see where Gemini 3.8 Live ranks.

Rating from head-to-head calls: on every call each pair of models is compared, the better call wins. 1000 is the average model. Higher is better.

#ModelELO Rating
  1. 01Grok Voice 3.01154
  2. 02GPT-Live 1 + GPT-6 Sol1069
  3. 03Gemini 3.1 Flash Live Preview1062
  4. 04GPT Realtime 2.11054
  5. 05Grok Voice Think Fast 2.01030
  6. 06Phonic-11030
  7. 07Azure Realtime1017
  8. 08Gemini 3.8 Live Extended Thinking1016
  9. 09Gemini 3.8 Live968
  10. 10GPT Realtime 2.1 Mini921
  11. 11Deepslate Opal847
  12. 12Amazon Nova 2 Sonic832
  • refCascade955

Where Gemini 3.8 Live Is Strong and Weak

Pick a skill family. The shape is this model on every skill in it; the dashed line is the median of the other models.

Gemini 3.8 LiveMedian of the other models
56Hearing Score9th of 12−11 vs median
SkillScore · vs median
  • Waiting Through Hold89%+5
  • Side-Speech Rejection100%+5
  • Multi-Speaker Handling43%0
  • Emotion Recognition82%−5
  • Degraded-Channel Robustness70%−5
  • Low-Confidence Confirmation10%−10
  • Entity Extraction Accuracy11%−27
  • Prosodic Cue Understanding40%−38

Share of calls where the skill was needed and handled, strongest first. Hover a point on the chart for both values.

Example Calls from Gemini 3.8 Live

Banking and healthcare callers. Each shows what the caller needed, what made the call hard, and how the model handled it.

First prenatal visit

Healthcare
Talks over the agent

Finished the task

Handled well: Overlapping-Speech Recovery, Redundant Re-prompt Avoidance, Hallucinated-Success Avoidance

Listen to the call

Urgent visit for a feverish child

Healthcare
Background voicesSlow systemAsks to holdNeeds consent

Finished the task

Handled well: Side-Speech Rejection, Mid-Utterance Pause Tolerance, Consent Verification

Struggled with: Tool-Latency Handling

Listen to the call

Reschedule a specialist visit

Healthcare
Bad connectionSeveral peopleTalks over the agentChanges their mind

Finished the task

Handled well: Degraded-Channel Robustness, Multi-Speaker Handling, Overlapping-Speech Recovery

Struggled with: Correction Handling, Consent Verification

Listen to the call

Update insurance details

Healthcare
Spells names and numbersBad connectionNeeds consentCancels

Finished 2 of 4 steps

Handled well: Degraded-Channel Robustness, Redundant Re-prompt Avoidance, Hallucinated-Success Avoidance

Struggled with: Entity Extraction Accuracy, Consent Verification, Cancellation Handling

Listen to the call

Loan payoff quote

Banking
Noisy lineHesitant callerAsks to holdNeeds consent

Finished 6 of 7 steps

Handled well: Prosodic Cue Understanding, Mid-Utterance Pause Tolerance, Consent Verification

Struggled with: Low-Confidence Confirmation, Numeric Read-back

Listen to the call

Change a mortgage payment date

Banking
Hold musicSeveral peopleInterruptsNeeds a person

Finished 5 of 7 steps

Handled well: Waiting Through Hold, Multi-Speaker Handling, Backchannel vs Barge-in Detection

Struggled with: Human Handoff

Listen to the call

Business loan documents

Banking
Slow systemAsks to holdNeeds consentAsks for a number

Finished 0 of 3 steps

Handled well: Mid-Utterance Pause Tolerance, Redundant Re-prompt Avoidance, Hallucinated-Success Avoidance

Struggled with: Tool-Latency Handling, Premature-Response Avoidance, Consent Verification

Listen to the call

Request a medical records transfer

Healthcare
Needs consentCancelsAsks for a number

Finished 3 of 7 steps

Handled well: Redundant Re-prompt Avoidance, Hallucinated-Success Avoidance, Numeric Read-back

Struggled with: Consent Verification, Cancellation Handling

Listen to the call

Payment plan for missed loan payments

Banking
Hesitant callerSeveral peoplePauses mid-sentenceNeeds consent

Finished 4 of 6 steps

Handled well: Prosodic Cue Understanding, Multi-Speaker Handling, Endpointing

Listen to the call

Remove a sold car from a policy

Banking
Noisy lineInterruptsChanges their mindSystem error

Finished 4 of 7 steps

Handled well: Backchannel vs Barge-in Detection, Correction Handling, Tool Error Handling

Struggled with: Low-Confidence Confirmation, Consent Verification

Listen to the call

Latency at P50, P90 and P95

Gemini 3.8 Live starts replying in 2.01s on a typical turn and 2.63s on the slowest turn in twenty, 5th fastest of 12 at the median.

ModelP50P90P95
  • Cascade
    2.06s2.49s2.60s
Seconds from the caller finishing to the agent replying. The bar is the typical turn, P50.

Gemini 3.8 Live Head to Head

Every comparisonAll 12 models

How Much Audio Gemini 3.8 Live Was Tested On

210 live calls: 70 scenarios, each run three times. The median call ran 3:02.

Live calls
210
Scenarios
70
Banking calls
105
Healthcare calls
105
Replies per call
13.5
Median call
3:02

70 scenarios, the same for every model.

Gemini 3.8 Live FAQ

  • Gemini 3.8 Live has an ELO rating of 968, 9th of 12 ranked models. It passed 19 of 70 callers on all three runs, a Pass³ of 27.1%.

  • Gemini 3.8 Live starts replying 2.01s after the caller finishes at P50, 2.52s at P90 and 2.63s at P95: 5th fastest of 12 at P50, measured on the call audio.

  • Gemini 3.8 Live falls furthest behind on Corrections and Changes: 37%, against a median of 61%.

  • Gemini 3.8 Live has a task completion rate of 16.2%, 9th of 12. A call counts only when every step the caller needed was done.

  • There is no verified per-minute rate for Gemini 3.8 Live, so it is left out of the cost ranking.

  • Cekura ran Gemini 3.8 Live as the whole agent with the same prompt, tools and phone connection as every other model. It took the same 70 recorded real human callers on live calls, three times each (210 calls), and each call was graded on 25 skills across Hearing, Turn Taking, Task Handling and Speaking.

Test your voice agent the same way.

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

Book a demoStart free