2 models compared on the same benchmark — accuracy, latency, price, language coverage and capabilities. Nova-2 leads on raw accuracy; Whisper Large V3 is fastest; Whisper Large V3 covers the most languages; Nova-2 is cheapest.
Verdict · who wins what
Most accurate
Nova-2
16.5% WER
Deepgram
Fastest
Whisper Large V3
0.3× realtime
OpenAI
Most languages
Whisper Large V3
98 languages
OpenAI
Cheapest
Nova-2
$0.0058/min
Deepgram
Comparing
2 models · same benchmark
1
Nova-2
Deepgram
Deepgram's Nova-2 speech recognition
2
Whisper Large V3
OpenAI
OpenAI's Whisper large-v3 model
30-day benchmark average
overall WER vs price
Nova-2
Whisper Large V3
field · 26 models
lower-left is better
WER · lower is better
WER · English
p50 → p99
Streaming benchmark averages
Rolling 30 days
billed per second
Feature support
Languages · formats · regions
general
legal
medical
noisy
technical
uk
us
Latency p99
3.7s
4.3s
Final drain
192ms
—
Real-time factor
1.02×
—
Flicker
10.5%
—
Cadence
0.8/s
—
$34.92
$36.00
1,000 hours
$349.20
$360.00
No
Auto-detect language
Yes
Yes
Live / streaming
Yes
No
Custom vocabulary
Yes
No
Compliance
Is Nova-2 or Whisper Large V3 more accurate?
Nova-2 is more accurate, with a 16.5% word error rate versus 17.7% for Whisper Large V3, on our standardized benchmark.
Which is cheaper, Nova-2 or Whisper Large V3?
Nova-2 is cheaper at $0.0058/min versus $0.0060/min for Whisper Large V3.
Should you choose Nova-2 or Whisper Large V3?
Choose Nova-2 for accuracy and lower cost; choose Whisper Large V3 for speed and language coverage.
