Cekura Bench
Voice AI benchmarks — compare leading voice agent platforms and speech models.
Public benchmarks comparing voice AI models and agent platforms on live phone calls, with open methodology and datasets.
Cekura Bench homepage
1 / 2Overview
What is Cekura Bench?
Voice AI benchmarks — compare leading voice agent platforms and speech models.
Cekura Bench publishes verifiable voice AI benchmarks: its new speech-to-speech benchmark tests voice AI models and agents on live phone calls, comparing voice agent platforms, realtime speech-to-speech models, production conversation quality, and speech-to-text models under documented test conditions.
A public leaderboard ranks platforms by pass^3 score with filters across 36 models and 17 platforms, plus dedicated head-to-head pages that compare models under identical conditions. Every benchmark documents its methodology and links to reports, test cases, and datasets (datasets licensed CC BY-NC).
The Bench is part of Cekura (YC 2024), which offers automated QA for voice and chat AI agents: pre-production simulation, production monitoring, and adversarial red-teaming. The benchmark site is free to browse; the platform starts at $0.25 per voice testing minute and $0.05 per monitored call, with a Startup plan at $500/mo and custom Enterprise plans.
Platforms and languages
Cekura Bench Availability
Platforms
Languages
Capabilities
Cekura Bench Key Features
Speech-to-speech benchmark
Live-phone-call tests comparing models such as GPT Realtime, ElevenLabs, and Deepgram.
Head-to-head comparisons
Dedicated pages comparing speech-to-speech and speech-to-text models under identical conditions.
Public leaderboard
Ranks platforms by pass^3 score with filters for 36 models and 17 platforms.
Open methodology and datasets
Each benchmark documents its methodology with links to reports, test cases, and datasets.
Broader Cekura platform
Pre-production simulation, production monitoring, adversarial red-teaming, and integrations with major voice AI platforms.
Best for
Who uses Cekura Bench?
Voice AI developers and enterprises
Evaluate and compare voice agents and speech models before building or buying
Plans and access
Cekura Bench Pricing
Freemium
Bench free to browse; platform $0.25/voice-test min, $0.05/monitored call; Startup $500/mo
Common questions
Cekura Bench FAQs
What does Cekura Bench compare?
Voice agent platforms, realtime speech-to-speech models, production conversation quality, and speech-to-text models — all under documented test conditions.
How are benchmark results measured?
By reliability, task completion, data accuracy, response time, voice quality, interruption handling, and speech recognition accuracy where applicable.
Can I compare two models directly?
Yes. Dedicated head-to-head pages compare speech-to-speech and speech-to-text models tested under the same conditions.
What does pass^3 mean?
The share of scenarios where all three retained test calls passed. A scenario must succeed repeatedly, not just once.
Can I inspect the methodology?
Yes. Each benchmark documents its methodology and links to reports, test cases, or source material.
Reviews
See what the community thinks and share your experience.
—
Based on 0 ratings
Rating distribution
Community reviews
No reviews
No written reviews yet.
