Canary Qwen 2.5B, NVIDIA
Jul 2025
Canary Qwen 2.5B, NVIDIACompare inon
Canary Qwen 2.5B, NVIDIA measured on 17 of 80 Speech to text benchmarks17
BenchmarkByScoreRank
No benchmark matches
$0.001per minute
4.3AA Speech to text
Jul 2025released
enLanguages
$0.001 / minPer audio minute
Exact native tariff; see source for billing scope.Price covers
Ranking?
Full ranking →···
37Gemma 4 E4B It4.139Canary Qwen 2.5B, NVIDIA4.140Gemma 4 12B It4.2Strong at
Alternatives?
ModelScoreCostPriceWhat changes
MCheaperMultilingualsame-32%$0.001
BetterVoxtral Small 24B 2507Open×2.1+305%$0.003Per audio minute: $0.001 / min → $0.003 / min+1
OpenVibeVoice ASR HF×2——Per audio minute: not verified+1
CheaperWhisper Large v3 TurboOpen-41%-10%$0.001Memory (4-bit): not verifiedABetterUniversal 3.5 Pro×2.2+373%$0.004Per audio minute: $0.001 / min → $0.004 / minModelScoreCostPriceWhat changes
MMultilingualsame-32%$0.001Whisper Large v3 TurboOpen-41%-10%$0.001Memory (4-bit): not verifiedModelScoreCostPriceWhat changes
Voxtral Small 24B 2507Open×2.1+305%$0.003Per audio minute: $0.001 / min → $0.003 / min+1AUniversal 3.5 Pro×2.2+373%$0.004Per audio minute: $0.001 / min → $0.004 / min
Scribe v2×2.9+396%$0.004Per audio minute: $0.001 / min → $0.004 / minOSpeech v3.1×2.9——Per audio minute: not verifiedAzure Speech 07 2026×2.5——Languages: not verified+1ModelScoreCostPriceWhat changes
VibeVoice ASR HFOpen×2——Per audio minute: not verified+1Voxtral Mini 4B Realtime 2602Open+59%——Per audio minute: not verified+1EARK ASR 0.6BOpen+50%——Per audio minute: not verified
Qwen 3 ASR 1.7B HfOpen+45%——Per audio minute: not verifiedQwen 3 ASR 0.6B HfOpen+43%——Per audio minute: not verifiedModelScoreCostPriceWhat changes
Canary 1B v2Opensame——Per audio minute: not verified+1Nemotron 3.5 ASR Streaming 0.6BOpensame——Per audio minute: not verified+1Parakeet TDT 0.6B V2, NVIDIAOpensame——Per audio minute: not verified+1Parakeet Ctc 1.1BOpensame——Per audio minute: not verified+1Parakeet TDT 0.6B V3, TogetheraiOpen+36%+103%$0.002Per audio minute: $0.001 / min → $0.002 / min+1Nemotron Speech Streaming En 0.6BOpen+72%——Languages: not verified+2Stt Hy Fastconformer Hybrid Large PcOpenn/a——Per audio minute: not verified+1Parakeet RNNT 1.1BOpenn/a—$0.002Per audio minute: $0.001 / min → $0.002 / min+1All NVIDIA models →Run it yourself
estimateFits on✓ 8 GB (phone, laptop) · full quality✓ 16 GB · full quality✓ 24 GB GPU · full quality✓ 48 GB GPU · full quality✓ 80 GB GPU · full quality
LicenceCC BY 4.0free for commercial use
Benchmarks
in Speech & VoiceOpen ASR · average WER6.4#25
Open ASR · RTFx881.7#19
Open ASR · AMI WER9.1#11
FAQ
How much does Canary Qwen 2.5B, NVIDIA cost?›
$0.001 per minute.
Is Canary Qwen 2.5B, NVIDIA free to use?›
There is no free API; it costs $0.001 per minute. The weights are open, so you can run it yourself.
How good is Canary Qwen 2.5B, NVIDIA, and what is it best at?›
#31 of 51 in Speech to text.
Is Canary Qwen 2.5B, NVIDIA better than Gemma 4 E4B It?›
On Open ASR · VoxPopuli AA Cleaned WER, Canary Qwen 2.5B, NVIDIA scores 4.1 vs 4.1 — about the same. It is 96% cheaper.
When was Canary Qwen 2.5B, NVIDIA released? Is there a newer version?›
Released 2025-07-17. It is the current version in its line.
Is Canary Qwen 2.5B, NVIDIA open source?›
Yes, the weights are open (cc-by-4.0 licence).
Can I run Canary Qwen 2.5B, NVIDIA locally?›
The weights are open and can be self-hosted; hardware needs are not confirmed yet.
What is Canary Qwen 2.5B, NVIDIA?›
Canary Qwen 2.5B, NVIDIA is an open-weights speech & voice model from NVIDIA, released Jul 2025.
What are the best alternatives to Canary Qwen 2.5B, NVIDIA?›
Multilingual (cheaper), Voxtral Small 24B 2507 (better), VibeVoice ASR HF (open weights).