A JOURNAL OF ARTIFICIAL INTELLIGENCE
my ai daily

ARTIFICIAL ANALYSIS

Word Error Rate.

Artificial Analysis non-streaming AA-WER index for speech-to-text models. Lower is better. Snapshot 29 Oct 2026.

Lower word error rate is better. AA-WER v2 is the non-streaming index. Speed factor is median input-audio seconds transcribed per second. Price is USD per 1,000 minutes.

Artificial Analysis speech-to-text leaderboard ↗Snapshot 29 Oct 2026. Rankings are Artificial Analysis’s, not scores from this newsroom.

RankModelProviderWhisperWord error rateMedian speed factorUSD / 1,000 min
1Fun-Realtime-ASR-preview, AlibabaAlibaba Cloud—1.7%20.6$2.82
2MAI-Transcribe-2Microsoft AI—2.0%310.4$1.67
3Voxtral Small Transcribe 3Mistral—2.1%64.2$4.00
4Scribe v2ElevenLabs—2.2%73.4$3.67
5Voxtral Mini Transcribe 3Mistral—2.2%76.6$3.50
6Grok Voice Transcribe 2.0SpaceXAI—2.3%182.9$1.67
7MAI-Transcribe-1.5Microsoft AI—2.4%81.3$6.00
8StepAudio 3 ASRStepFun—2.5%64.0$6.67
9Gemini 3.5 TranscribeGoogle—2.6%89.0$5.00
10Zoom Scribe V2 ProZoom—2.8%124.0$3.33
11Voxtral SmallMistral—2.8%57.3$4.00
12Gemini 3.1 Pro Preview (High)Google—2.8%8.1$18.15
13Gemini 2.5 ProGoogle—2.9%9.7$11.39
14Gemini 3 Flash (High)Google—2.9%21.0$13.70
15Universal-3.5 ProAssemblyAI—3.0%104.2$3.50
16Universal-3 ProAssemblyAI—3.1%100.0$3.50
17Solaria-3Gladia—3.2%92.9$10.16
18GPT TranscribeOpenAI—3.3%46.1$4.50
19Smallest AI PulseSmallest.ai—3.3%254.8$5.00
20Resonant-1Reson8—3.4%337.1$3.60
21Smallest AI Pulse ProSmallest.ai—3.4%297.6$4.00
22Gemini 3.1 Flash-Lite Preview (Minimal)Google—3.4%89.9$5.83
23Qwen3.5 Omni PlusAlibaba Cloud—3.5%149.1$0.00
24Gemini 3.1 Pro Preview (Low)Google—3.6%8.5$7.72
25Voxtral Mini Transcribe 2Mistral—3.6%86.7$3.00
26Soniox v5 AsyncSoniox—3.8%41.0$1.66
27UniversalAssemblyAI—3.8%102.8$2.50
28Voxtral Mini, DeepinfraDeepInfra—3.8%47.1$1.00
29Inworld STT 1Inworld—3.9%142.0$2.50
30GPT-4o TranscribeOpenAI—4.0%37.9$6.00
31EnhancedSpeechmatics—4.0%52.3$12.50
32Large v2OpenAIlarge-v24.1%28.5$6.00
33Solaria-1Gladia—4.1%92.2$10.17
34Whisper (L, v3), fal.aifal.ailarge-v34.1%91.5$1.15
35Amazon TranscribeAmazon Bedrock—4.1%34.9$6.00
36Modulate STT Batch English VFastModulate—4.2%55.5$0.42
37Canary Qwen 2.5B, ReplicateReplicate—4.3%8.4$0.74
38Chirp 3Google—4.3%29.6$16.00
39GPT-4o Mini TranscribeOpenAI—4.5%45.4$3.00
40Whisper Large v3, together.aiTogether AIlarge-v34.5%294.4$1.50
41Parakeet TDT 0.6B V3, TogetheraiTogether AI—4.5%687.0$1.50
42Cohere Transcribe 03-2026Cohere—4.6%66.3$0.00
43Whisper (L, v3, Turbo), GroqGroqv3 Turbo4.6%276.8$0.67
44StepAudio 2.5 ASRStepFun—4.7%115.3$0.37
45Wizper (L, v3), fal.aifal.ailarge-v34.7%152.3$0.50
46Nova-3, TelnyxTelnyx—4.8%500.9$7.40
47Nova 2 ProAmazon Bedrock—4.9%23.4$3.10
48MeliaSpeechmatics—4.9%176.8$4.00
49Nova-2, TelnyxTelnyx—5.1%457.5$7.40
50StandardSpeechmatics—5.1%87.8$7.50
51Gemini 2.5 FlashGoogle—5.1%66.7$6.66
52Gemini 2.5 Flash LiteGoogle—5.2%74.6$6.56
53Nova-3Deepgram—5.2%623.2$4.30
54Parakeet RNNT 1.1B, ReplicateReplicate—5.4%6.3$1.91
55Whisper Large v3 Turbo, TelnyxTelnyx—5.5%37.0$15.00
56Incredibly Fast, ReplicateReplicatelarge-v35.7%55.0$1.49
57Rev AIRev AI—5.9%13.1$3.33
58Parakeet TDT 0.6B V2NVIDIA—6.4%97.4$0.00
59Gradium Speech-to-TextGradium—6.8%2.3$13.00
60Whisper (L, v3), ReplicateReplicatelarge-v310.1%2.9$4.23
61Qwen3.5 Omni FlashAlibaba Cloud—13.5%70.9$0.00
62ChirpGoogle—31.2%14.3$16.00

Follow your curiosity.

98 STORIES TO EXPLORESOURCES & CONTEXT INCLUDED