Skip to main content

ElevenLabsDeepgramAudio matchupSpeech to text

Scribe v2 Realtimevs.Nova-3

Which model is better?

Scribe v2 Realtime is slightly ahead in independent tests, but the lead is not certain.

Independent tests
Scribe v2 Realtime0 of 1
Nova-30 of 1
AirCaps A5Sv2 ASR benchmark
Popularity
Scribe v2 Realtime29
Nova-3ahead51
out of 100
Weights and access
Scribe v2 Realtimeproprietary
Nova-3proprietary
Released
Scribe v2 Realtimen/a
Nova-3n/a

Pick Scribe v2 Realtime for

leaning AirCaps A5Sv2 ASR benchmark

Pick Nova-3 for

more popular

Shared tests

Independent evaluators and leaderboards only, the newest shared snapshot per evaluator. With 95 % ranges a lead counts when they do not overlap; for hit rates, when it holds in a 95 % test. Without any uncertainty figure at most “slightly ahead”.

AirCaps A5Sv2 ASR benchmark

Scribe v2 Realtime slightly ahead

Scribe v2 Realtime is clearly ahead in 0 of 6 subtests and slightly in 6, Nova-3 in none.

AirCaps A5Sv2 DiPCo corpus WER

Scribe v2 Realtime31.7 %
Nova-368.2 %

6 tasks, lower is better, as of October 19, 2026

AirCaps A5Sv2 Mega-ASR corpus WER

Scribe v2 Realtime22.6 %
Nova-340.5 %

1,250 tasks, lower is better, as of October 19, 2026

AirCaps A5Sv2 AMI corpus WER

Scribe v2 Realtime20.4 %
Nova-335.9 %

7 tasks, lower is better, as of October 19, 2026

Show all 6 subtests
SubtestScribe v2 RealtimeNova-3Verdict
AirCaps A5Sv2 DiPCo corpus WER (lower is better)31.7 %68.2 %Scribe v2 Realtime slightly ahead
AirCaps A5Sv2 Mega-ASR corpus WER (lower is better)22.6 %40.5 %Scribe v2 Realtime slightly ahead
AirCaps A5Sv2 AMI corpus WER (lower is better)20.4 %35.9 %Scribe v2 Realtime slightly ahead
AirCaps A5Sv2 Pooled WER (source-published word-pooled aggregate) (lower is better)26.7 %45.7 %Scribe v2 Realtime slightly ahead
AirCaps A5Sv2 Macro WER (source-published four-corpus mean) (lower is better)26.6 %45.6 %Scribe v2 Realtime slightly ahead
AirCaps A5Sv2 NOTSOFAR corpus WER (lower is better)31.8 %37.7 %Scribe v2 Realtime slightly ahead

Source · latest October 19, 2026

All data side by side

Prices are verbatim from the provider, because audio models bill per minute, character or token.

Scribe v2 RealtimeNova-3
TaskSpeech to textSpeech to text
DeveloperElevenLabsDeepgram
StatusActiveActive
Weightsproprietaryproprietary
Runs locallynono
Popularity (out of 100)2951