Leaderboards
Benchmarks
Every score carries its origin.
Accented speech76Clean read speech92European-accented speech92Financial calls92Harder read speech92Open ASR RTFx74Open ASR WER76Podcasts and video92Recorded meetings92
Recorded meetings
Word error rate on AMI — real meetings recorded in a room, with people talking over each other and a microphone that is not close — LOWER is betterProvenance92 modelsscores Aug 2, 2026 – Sep 28, 2026
1Scribe v2 ProZoomScore6.08ScoredSep 11, 20262Modulate MultilingualModulateScore6.31ScoredSep 11, 20263Modulate VfastModulateScore6.72ScoredAug 2, 20264Scribe v1ZoomScore6.94ScoredAug 2, 20265Azure Speech 06 2026MicrosoftScore6.95ScoredAug 2, 20266Granite Speech 4.1 2b NARIBM2.3BScore6.96ScoredAug 2, 20267Cohere Transcribe 03 2026Cohere Labs2.1BScore7.02ScoredSep 11, 20268Azure Speech 07 2026MicrosoftScore7.05ScoredSep 14, 20269Granite Speech 4.1 2bIBM2.3BScore7.06ScoredAug 2, 202610Nyra Health CrisperWhisperNyra Health1.6BScore7.1ScoredAug 2, 202611Granite Speech 5.0 470m Turboctc ncIBM0.5BScore7.13ScoredSep 11, 202612Higgs Audio v3 STTBoson AI2.7BScore7.19ScoredAug 2, 202613ASR K1 (preview)sopheaScore7.29ScoredSep 14, 202614Granite 4.0 1b SpeechIBM2.3BScore7.38ScoredSep 11, 202615Hojo ASR V1Hojo AI5.2BScore7.54ScoredSep 11, 202616Granite Speech 3.3 8bIBM8.6BScore7.7ScoredAug 2, 202617Granite Speech 5.0 470m TurboctcIBM0.5BScore7.72ScoredSep 11, 202618Granite Speech 3.3 2bIBM3BScore7.8ScoredAug 2, 202619MOSS Transcribe Preview 2BOpenMOSS2.4BScore7.8ScoredAug 2, 202620ARK ASR 3BAutoArk AI4.1BScore7.89ScoredAug 2, 202621Canary Qwen 2.5bNVIDIA2.6BScore7.91ScoredAug 2, 202622Scribe v2ElevenLabsScore8.19ScoredSep 11, 202623Resonant 1 FlashReson8Score8.21ScoredAug 2, 202624Resonant 1Reson8Score8.22ScoredAug 2, 202625Qwen3 ASR 1.7B HFQwen2BScore8.31ScoredSep 11, 202626Qwen3 ASR 1.7BQwen2.3BScore8.31ScoredAug 2, 202627MOSS Transcribe DiarizeOpenMOSS0.9BScore8.33ScoredSep 11, 202628Higgs Audio v3 8b STT v2Boson AI8.9BScore8.37ScoredAug 2, 202629ARK ASR 0.6BAutoArk AI1.3BScore8.61ScoredSep 11, 202630Nemotron Speech Streaming en 0.6bNVIDIA0.6BScore8.84ScoredAug 2, 202631Moonshine Streaming MediumUseful Sensors0.3BScore8.93ScoredAug 2, 202632Parakeet TDT 0.6b v3 (fast Gpu Asr)NVIDIA0.6BScore9.02ScoredSep 28, 202633sprag SymphonyspragScore9.02ScoredSep 28, 202634Parakeet TDT 0.6b v2 (fast Gpu Asr)NVIDIA0.6BScore9.02ScoredSep 28, 202635Parakeet TDT 0.6b v2NVIDIA0.6BScore9.09ScoredSep 11, 202636Smallest AI PulseSmallest AIScore9.11ScoredSep 11, 202637Phi 4 Multimodal InstructMicrosoft5.6BScore9.12ScoredSep 11, 202638Muse Voice TranscribeMetaScore9.2ScoredSep 21, 202639Qwen3 ASR 0.6B HFQwen0.8BScore9.33ScoredSep 11, 202640Parakeet TDT 0.6b v3NVIDIA0.6BScore9.42ScoredSep 11, 202641Qwen3 ASR 0.6BQwen0.9BScore9.47ScoredAug 2, 202642Avalon v1 enAqua VoiceScore9.71ScoredAug 2, 202643Solaria 3GladiaScore9.89ScoredAug 2, 202644Speechmatics EnhancedSpeechmaticsScore9.99ScoredSep 11, 202645Zipformer cr CTC Transducer XL 290M (fast Gpu Asr, transducer Modified Beam Search)Sounds Good AI0.3BScore10.24ScoredSep 28, 202646Zipformer cr CTC Transducer XL 290MSounds Good AI0.3BScore10.28ScoredAug 2, 202647Fun ASR Nano 2512 HFFunAudioLLM0.8BScore10.35ScoredSep 14, 202648STT 2.6b enKyutai2.6BScore10.52ScoredSep 11, 202649Canary 1b FlashNVIDIA0.8BScore10.58ScoredAug 2, 202650Universal 3 5 ProAssemblyAIScore10.63ScoredAug 2, 202651Moonshine Streaming SmallUseful Sensors0.1BScore10.79ScoredAug 2, 202652Canary 1bNVIDIA1BScore10.87ScoredAug 2, 202653Audio8 ASR 0.1BAutoArk AI0.3BScore11ScoredSep 11, 202654Universal 3 ProAssemblyAIScore11.06ScoredAug 2, 202655Owsm CTC v4 1BESPnet1BScore11.1ScoredSep 11, 202656Zipformer cr CTC Transducer XL 290M (fast Gpu Asr, ctc Greedy Search)Sounds Good AI0.3BScore11.64ScoredSep 28, 202657Niagara 84m Batch.enApplied Brain Research84MScore12.01ScoredSep 28, 202658VibeVoice ASR HFMicrosoft8.3BScore12.02ScoredSep 11, 202659Canary 180m FlashNVIDIA0.2BScore12.09ScoredSep 11, 202660Distil Large v3.5Distil-Whisper0.8BScore12.09ScoredAug 2, 202661Parakeet CTC 1.1bNVIDIA1.1BScore12.23ScoredSep 11, 202662Parakeet TDT CTC 110mNVIDIA0.1BScore12.69ScoredAug 2, 202663Zipformer Transducer XL 290MSounds Good AI0.3BScore12.8ScoredAug 2, 202664Parakeet TDT 1.1bNVIDIA1.1BScore12.87ScoredAug 2, 202665Canary 1b v2NVIDIA1BScore13.03ScoredSep 11, 202666Parakeet CTC 0.6bNVIDIA0.6BScore13.09ScoredSep 11, 202667Voxtral Small 24B 2507Mistral AI24.3BScore13.19ScoredSep 11, 202668Voxtral Mini 4B Realtime 2602Mistral AI4.4BScore13.34ScoredAug 2, 202669Owsm CTC v3.1 1BESPnet1.1BScore13.35ScoredAug 2, 202670Nemotron 3.5 ASR Streaming 0.6bNVIDIA0.6BScore13.43ScoredSep 11, 202671Voxtral Mini 3B 2507Mistral AI4.7BScore13.57ScoredSep 11, 202672Whisper Large v3OpenAI1.5BScore13.63ScoredAug 2, 202673Lite Whisper Large v3 AccEfficient Speech1.4BScore13.69ScoredAug 2, 202674omniASR LLM 7B v2Meta7.8BScore13.86ScoredAug 2, 202675Niagara 38m Batch.enApplied Brain Research38MScore13.87ScoredSep 11, 202676Owsm CTC v3.2 ft 1BESPnet1BScore13.88ScoredAug 2, 202677Whisper Large v3 TurboOpenAI0.8BScore13.88ScoredSep 11, 202678GLM ASR Nano 2512Z.ai2.3BScore13.95ScoredAug 2, 202679Parakeet RNNT 1.1bNVIDIA1.1BScore14.02ScoredAug 2, 202680Parakeet RNNT 0.6bNVIDIA0.6BScore14.34ScoredAug 2, 202681ASR Conformer LargescaleasrSpeechBrain0.5BScore15.57ScoredAug 2, 202682Niagara 19m Batch.enApplied Brain Research20MScore16.17ScoredSep 28, 202683Moonshine Streaming TinyUseful Sensors0Score16.95ScoredSep 11, 202684omniASR CTC 7B v2Meta6.5BScore17.44ScoredAug 2, 202685Moonshine TinyUseful Sensors0Score18.82ScoredSep 11, 202686Gemma 4 E4B itGoogle8BScore20.57ScoredSep 14, 202687Niagara 9m Batch.enApplied Brain Research9MScore20.94ScoredSep 28, 202688Gemma 4 E2B itGoogle5.1BScore24.77ScoredSep 14, 202689Wav2vec2 Large 960h Lv60 SelfMeta0.3BScore25.83ScoredAug 2, 202690Hubert Large Ls960 ftMeta0.3BScore28.9ScoredAug 2, 202691MMS 1b AllMeta1BScore31.24ScoredAug 2, 202692Gemma 4 12B itGoogle12BScore75.79ScoredSep 14, 2026
