Speech to Text AI Model & Provider Leaderboard
Compare word error rate, speed, and pricing across Speech to Text models and providers.
For further details, see our methodology page.
Artificial Analysis Word Error Rate Index (Non-streaming)
Artificial Analysis Word Error Rate Index (Non-streaming)
AA-WER (Non-streaming) by Dataset
AA-WER (Non-streaming): AA-AgentTalk Dataset
Cleaned Dataset Comparison
VoxPopuli: Cleaned vs Original Subset of Publicly Available Data
API Benchmarks
Artificial Analysis Word Error Rate Index (Non-streaming) vs. Price
Speed Factor
Price of Transcription
Summary of Key Metrics & Further Information
Provider | Further Details | ||||
|---|---|---|---|---|---|
Qwen3.5 Omni Flash |  Alibaba Cloud | 13.5% | 70.9 | 0.00 | |
Qwen3.5 Omni Plus | Alibaba Cloud | 3.5% | 148.4 | 0.00 | |
Fun-Realtime-ASR-preview | Alibaba Cloud | 1.7% | 20.6 | 2.82 | |
Nova 2 Pro |  Amazon Bedrock | 4.9% | 23.5 | 3.10 | |
Amazon Transcribe | Amazon Bedrock | 4.1% | 34.3 | 6.00 | |
Universal-3.5 Pro |  AssemblyAI | 3.0% | 101.5 | 3.50 | |
Universal-3 Pro | AssemblyAI | 3.1% | 100.0 | 3.50 | |
Universal, AssemblyAI | AssemblyAI | 3.8% | 102.4 | 2.50 | |
MAI-Transcribe-2 |  Microsoft AI | 2.0% | 310.3 | 1.67 | |
MAI-Transcribe-1.5 | Microsoft AI | 2.4% | 81.3 | 6.00 | |
transcribe-03-2026 |  Cohere | 4.6% | 66.1 | 0.00 | |
Nova-3 |  Deepgram | 5.2% | 623.2 | 4.30 | |
Nova-3, Telnyx |  Telnyx | 4.8% | 497.5 | 7.40 | |
Nova-2, Telnyx | Telnyx | 5.1% | 457.5 | 7.40 | |
Scribe v2 |  ElevenLabs | 2.2% | 72.2 | 3.67 | |
Solaria-1, Gladia |  Gladia | 4.1% | 92.2 | 10.17 | |
Solaria-3, Gladia | Gladia | 3.2% | 92.3 | 10.16 | |
Chirp 3, Google |  Google | 4.3% | 29.6 | 16.00 | |
Chirp | Google | 31.2% | 14.3 | 16.00 | |
Gemini 3.5 Transcribe | Google | 2.6% | 89.0 | 5.00 | |
Gemini 3.1 Pro Preview (High) | Google | 2.8% | 8.2 | 18.15 | |
Gemini 3.1 Pro Preview (Low) | Google | 3.6% | 8.6 | 7.72 | |
Gemini 3 Flash (High) | Google | 2.9% | 21.9 | 13.70 | |
Gemini 2.5 Flash Lite | Google | 5.2% | 74.6 | 6.56 | |
Gemini 2.5 Flash | Google | 5.1% | 66.1 | 6.66 | |
Gemini 2.5 Pro | Google | 2.9% | 9.7 | 11.39 | |
Gemini 3.1 Flash-Lite Preview (Minimal) | Google | 3.4% | 89.9 | 5.83 | |
Gradium Speech-to-Text |  Gradium | 6.8% | 2.3 | 13.00 | |
Grok Voice Transcribe 2.0 |  SpaceXAI | 2.3% | 184.4 | 1.67 | |
Inworld STT 1 |  Inworld | 3.9% | 142.0 | 2.50 | |
Voxtral Small Transcribe 3 |  Mistral | 2.1% | 63.2 | 4.00 | |
Voxtral Mini Transcribe 3 | Mistral | 2.2% | 76.6 | 3.50 | |
Voxtral Mini Transcribe 2 | Mistral | 3.6% | 85.2 | 3.00 | |
Voxtral Small | Mistral | 2.8% | 57.3 | 4.00 | |
Voxtral Mini |  DeepInfra | 3.8% | 47.1 | 1.00 | |
Modulate STT Batch English VFast |  Modulate | 4.2% | 55.1 | 0.42 | |
Nemotron 3.5 Transcribe | DeepInfra | 3.4% | 73.8 | 2.00 | |
Parakeet TDT 0.6B V3, Togetherai |  Together AI | 4.5% | 675.6 | 1.50 | |
Canary Qwen 2.5B, NVIDIA |  Replicate | 4.3% | 8.4 | 0.74 | |
Parakeet TDT 0.6B V2, NVIDIA |  NVIDIA | 6.4% | 95.9 | 0.00 | |
Parakeet RNNT 1.1B | Replicate | 5.4% | 6.3 | 1.91 | |
GPT Transcribe, OpenAI |  OpenAI | 3.3% | 45.0 | 4.50 | |
GPT-4o Transcribe | OpenAI | 4.0% | 37.8 | 6.00 | |
GPT-4o Mini Transcribe | OpenAI | 4.5% | 45.4 | 3.00 | |
Smallest AI Pulse Pro |  Smallest.ai | 3.4% | 296.3 | 4.00 | |
Resonant-1 |  Reson8 | 3.4% | 337.1 | 3.60 | |
Rev AI |  Rev AI | 5.9% | 13.1 | 3.33 | |
Smallest AI Pulse | Smallest.ai | 3.3% | 259.6 | 5.00 | |
Soniox v5 Async |  Soniox | 3.8% | 41.0 | 1.66 | |
Speechmatics Melia |  Speechmatics | 4.9% | 178.3 | 4.00 | |
Speechmatics Standard | Speechmatics | 5.1% | 87.8 | 7.50 | |
Speechmatics Enhanced | Speechmatics | 4.0% | 52.3 | 12.50 | |
StepAudio 2.5 ASR, StepFun |  StepFun | 4.7% | 115.3 | 0.37 | |
StepAudio 3 ASR, StepFun | StepFun | 2.5% | 64.0 | 6.67 | |
Whisper Large v3 Turbo |  Groq | 4.6% | 271.9 | 0.67 | |
Whisper Large v3 Turbo, Telnyx | Telnyx | 5.5% | 37.0 | 15.00 | |
Wizper Large v3 |  fal.ai | 4.7% | 152.1 | 0.50 | |
Incredibly Fast Whisper | Replicate | 5.7% | 55.7 | 1.49 | |
Whisper Large v3 | Replicate | 10.1% | 2.9 | 4.23 | |
Whisper Large v3 | fal.ai | 4.1% | 94.1 | 1.15 | |
Whisper Large v3 | Together AI | 4.5% | 294.4 | 1.50 | |
Whisper Large v2 | OpenAI | 4.1% | 28.5 | 6.00 | |
Zoom Scribe V2 Pro |  Zoom | 2.8% | 124.0 | 3.33 |
Frequently Asked Questions
Fun-Realtime-ASR-preview leads with the lowest AA-WER (Artificial Analysis Word Error Rate) of 1.7% across 62 models evaluated.
The top speech to text models by accuracy (AA-WER) are: 1. Fun-Realtime-ASR-preview (1.7%), 2. MAI-Transcribe-2 (2.0%), 3. Voxtral Small Transcribe 3 (2.1%), 4. Scribe v2, ElevenLabs (2.2%), and 5. Voxtral Mini Transcribe 3 (2.2%). Lower AA-WER indicates better transcription accuracy.
Parakeet TDT 0.6B V3, Togetherai is the fastest with a speed factor of 675.6x real-time, followed by Nova-3 (623.2x) and Nova-3, Telnyx (497.5x). Higher speed factors mean faster transcription.
StepAudio 2.5 ASR is the most affordable at $0.3667 per 1,000 minutes, followed by Modulate STT Batch English VFast ($0.417) and Wizper (L, v3), fal.ai ($0.50).
Voxtral Small, Mistral is the most accurate open weights model with an AA-WER of 2.8%. There are 15 open weights models out of 62 total evaluated.
The top open weights speech to text models by accuracy are: 1. Voxtral Small, Mistral (AA-WER 2.8%), 2. Nemotron 3.5 Transcribe (AA-WER 2.8%), and 3. Inkling (256K), Thinking Machines (AA-WER 3.5%).
The best model depends on your priorities. Use the scatter plots to visualize trade-offs between accuracy (AA-WER), speed, and price. For applications requiring high accuracy, prioritize models with lower AA-WER scores. For real-time applications, focus on speed factor. For cost-sensitive workloads, compare the price charts.