Deepslate Opal

847

ELO

11th of 12

12.9%

Pass³

11th of 12

2.46s

Latency

10th of 12

$0.112

Cost/Min

8th of 8

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)

Deepslate Opal Against Every Other Model

Each model took the same recorded human callers on live calls. Pick a measure to see where Deepslate Opal 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 Deepslate Opal 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.

Deepslate OpalMedian of the other models
60Hearing Score8th of 12−7 vs median
SkillScore · vs median
  • Multi-Speaker Handling50%+7
  • Waiting Through Hold88%+4
  • Entity Extraction Accuracy32%−5
  • Emotion Recognition81%−6
  • Degraded-Channel Robustness68%−7
  • Side-Speech Rejection89%−8
  • Low-Confidence Confirmation6%−14
  • Prosodic Cue Understanding63%−15

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

Example Calls from Deepslate Opal

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 systemNeeds consentAsks for a number

Finished 3 of 4 steps

Handled well: Side-Speech Rejection, Redundant Re-prompt Avoidance, Hallucinated-Success Avoidance

Struggled with: Tool-Latency Handling, Consent Verification

Listen to the call

Reschedule a specialist visit

Healthcare
Bad connectionSeveral peopleTalks over the agentChanges their mind

Finished 6 of 7 steps

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

Listen to the call

Update insurance details

Healthcare
Spells names and numbersBad connectionNeeds consentCancels

Finished 4 of 6 steps

Handled well: Entity Extraction Accuracy, Degraded-Channel Robustness, Redundant Re-prompt Avoidance

Struggled with: Consent Verification, Cancellation Handling

Listen to the call

Loan payoff quote

Banking
Noisy lineHesitant callerAsks to holdNeeds consent

Finished 4 of 7 steps

Handled well: Prosodic Cue Understanding, Mid-Utterance Pause Tolerance, Redundant Re-prompt Avoidance

Struggled with: Low-Confidence Confirmation, Consent Verification, Cancellation Handling

Listen to the call

Change a mortgage payment date

Banking
Hold musicSeveral peopleInterruptsNeeds a person

Finished 4 of 7 steps

Handled well: Waiting Through Hold, Multi-Speaker Handling, Redundant Re-prompt Avoidance

Struggled with: Backchannel vs Barge-in Detection, Human Handoff

Listen to the call

Business loan documents

Banking
Slow systemAsks to holdNeeds consent

Finished 0 of 2 steps

Handled well: Mid-Utterance Pause Tolerance, Hallucinated-Success Avoidance, TTS Text Normalization

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

Listen to the call

Request a medical records transfer

Healthcare
Slow systemNeeds consentAsks for a number

Finished 1 of 5 steps

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

Struggled with: Tool-Latency Handling, Consent Verification

Listen to the call

Payment plan for missed loan payments

Banking
Hesitant callerSeveral peoplePauses mid-sentenceNeeds consent

Finished 3 of 5 steps

Handled well: Prosodic Cue Understanding, Multi-Speaker Handling, Premature-Response Avoidance

Struggled with: Endpointing

Listen to the call

Remove a sold car from a policy

Banking
Noisy lineInterruptsChanges their mindSystem error

Finished 3 of 7 steps

Handled well: Tool Error Handling, Redundant Re-prompt Avoidance, Hallucinated-Success Avoidance

Struggled with: Low-Confidence Confirmation, Backchannel vs Barge-in Detection, Correction Handling

Listen to the call

Latency at P50, P90 and P95

Deepslate Opal starts replying in 2.46s on a typical turn and 3.36s on the slowest turn in twenty, 10th 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.

How Much Audio Deepslate Opal Was Tested On

210 live calls: 70 scenarios, each run three times. The median call ran 5:13.

Live calls
210
Scenarios
70
Banking calls
105
Healthcare calls
105
Replies per call
24.1
Median call
5:13

70 scenarios, the same for every model.

Deepslate Opal FAQ

  • Deepslate Opal has an ELO rating of 847, 11th of 12 ranked models. It passed 9 of 70 callers on all three runs, a Pass³ of 12.9%.

  • Deepslate Opal starts replying 2.46s after the caller finishes at P50, 3.16s at P90 and 3.36s at P95: 10th fastest of 12 at P50, measured on the call audio.

  • Deepslate Opal falls furthest behind on Barge-in and Overlap: 48%, against a median of 98%.

  • Deepslate Opal has a task completion rate of 12.4%, 12th of 12. A call counts only when every step the caller needed was done.

  • $0.112 per minute of call at Deepslate's published rates, 8th cheapest of 8 priced models. The vendor's regular rate of €0.10 a minute, at the ECB rate of 5 Oct 2026, applied to each call's length; its launch offer is €0.02. Excludes the ElevenLabs voice, billed on the customer's own ElevenLabs key.

  • Cekura ran Deepslate Opal 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.