Speech-to-text - verified
Speech-to-text API pricing by provider
39 speech-to-text models from 12 providers, read on October 18, 2026. Each price is shown in the provider's own unit. For the 35 priced per second, minute or hour we add a converted dollars-per-hour figure so you can sort them; it is our arithmetic and is labelled as such. 18 of the 39 are documented as streaming.
Every speech-to-text model
Price is the provider's own figure, copied as stated. Converted $/hour is our arithmetic: per second times 3,600, per minute times 60, per hour as is, with the stated rate under it. Token-priced models are not converted. Open a model for its quotes and sources.
39 models
| Model | Priceprovider's own unit | Converted $/hourour arithmetic | Streaming | Languages | Max audio | Features | Status |
|---|---|---|---|---|---|---|---|
| No models match these filters. | |||||||
GPT-TranscribeOpenAI - gpt-transcribe | $0.0045 / minutedevelopers.openai.com |
| Yes | Not stated | 25 MB per file |
| current (no preview label) |
GPT-Live-TranscribeOpenAI - gpt-live-transcribe | $0.017 / minutedevelopers.openai.com |
| Yes | Not stated | Not stated |
| current (no preview label) |
GPT-Realtime-WhisperOpenAI - gpt-realtime-whisper | $0.017 / minutedevelopers.openai.com |
| Yes | Not stated | Not stated | None recorded | current (no preview label) |
GPT-4o mini TranscribeOpenAI - gpt-4o-mini-transcribe | $0.003 / minute ($1.25 input / $5.00 output per 1M tokens)developers.openai.com |
| Yes | Not stated | 25 MB per file | None recorded | Retiring 2027-02-26deprecated (shutdown 2027-02-26) |
Gemini 3.5 TranscribeGemini API - gemini-3.5-transcribe | $2.00 per 1M audio input tokens ($0.003/min); $12.00 per 1M text output tokens ($0.002/min)ai.google.dev | Not convertedPriced per token | No | 85+ languages | 1 hour per request (30 minutes with diarization or word timestamps) |
| stable (no shutdown date) |
Gemini 3.5 Transcribe LiveGemini API - gemini-3.5-transcribe-live | $3.50 per 1M audio input tokens ($0.005/min); $21.00 per 1M text output tokens ($0.004/min)ai.google.dev | Not convertedPriced per token | Yes | 85+ languages | 10 minutes per session |
| stable (no shutdown date) |
Cloud Speech-to-Text V2, Chirp 3 (Standard recognition)Google Cloud - chirp_3 | Not stated | Not convertedNo price stated | Yes | 111 locales (29 GA, 82 Preview) in the transcription language tablederived | BatchRecognize: 1 minute to 1 hour (20 minutes with word-level timestamps) |
| GA (us and eu multi-regions) |
Cloud Speech-to-Text V2 dynamic batch (Standard models)Google Cloud - Speech-to-Text V2 dynamic batch | $0.003 / 1 minutecloud.google.com |
| No | Not stated | Not stated | None recorded | Not stated |
Amazon Transcribe standard batchAmazon - Amazon Transcribe (batch) | $0.006 per minute (US East (N. Virginia))aws.amazon.com - see note |
| No | 113 language codes in the supported-languages tablederived | 28,800 seconds and 2 GB per file |
| Not stated |
Amazon Transcribe standard streamingAmazon - Amazon Transcribe (streaming) | $0.01 per minute (US East (N. Virginia))aws.amazon.com - see note |
| Yes | 99 of the 113 language codes in the supported-languages table support streamingderived | Not stated |
| Not stated |
Amazon Transcribe MedicalAmazon - Amazon Transcribe Medical | $0.075 per minute (batch and streaming)aws.amazon.com - see note |
| Not stated | Not stated | Not stated | None recorded | Not stated |
Azure Speech real-time speech to textAzure - Standard Transcription - Real-time | $1 per hourazure.microsoft.com |
| Yes | Not stated | 240 minutes per session with real-time diarization |
| Not stated |
Azure Speech batch transcriptionAzure - Standard Transcription - Batch | $0.18 per hourazure.microsoft.com |
| No | Not stated | 1 GB per file |
| Not stated |
Azure Speech fast transcriptionAzure - Fast Transcription | $0.36 per 1 Hour (East US retail price)from Azure's price APIprices.azure.com - see note |
| No | Not stated | < 500 MB and < 5 hours per file | None recorded | Not stated |
MAI-Transcribe-2 (via Azure Speech)Azure - MAI-Transcribe-2 | $0.54 per 1 Hour (East US retail price)from Azure's price APIprices.azure.com - see note |
| Not stated | Not stated | Not stated | None recorded | Not stated |
Scribe v2ElevenLabs - scribe_v2 | $0.22 per hourelevenlabs.io |
| Not stated | 90+ languages | 3 GB; up to 10 hours (1 hour multichannel) |
| Not stated |
Scribe v2 RealtimeElevenLabs - scribe_v2_realtime | $0.39 per hourelevenlabs.io |
| Yes | 90+ languages | Not stated |
| Not stated |
Scribe v2 MedicalElevenLabs - scribe_v2_medical | $0.22 per hourelevenlabs.io |
| No | 90+ languages | Not stated |
| Not stated |
Flux EnglishDeepgram - flux-general-en | $0.0077/min streaming (regular price)deepgram.com - see note |
| Yes | English | Not stated |
| Not stated |
Flux MultilingualDeepgram - flux-general-multi | $0.0078/min streamingdeepgram.com |
| Yes | 10 languagesderived | Not stated | None recorded | Not stated |
Nova-3 MonolingualDeepgram - nova-3 | $0.0043/min pre-recorded; $0.0077/min streaming (regular price)deepgram.com - see note |
| Yes | 45+ languages (Nova models) | Not stated |
| Not stated |
Nova-3 MultilingualDeepgram - nova-3 (language=multi) | $0.0052/min pre-recorded; $0.0092/min streaming (regular price)deepgram.com - see note |
| Yes | 10 languages with code-switchingderived | Not stated | None recorded | Not stated |
Deepgram Whisper Cloud (Whisper Large)Deepgram - whisper-large | $0.0048/min pre-recordeddeepgram.com |
| No | Not stated | up to 20 minutes of processing time | None recorded | Not stated |
Universal-3.5 Pro (pre-recorded)AssemblyAI - Universal-3.5 Pro | $0.21/hrassemblyai.com |
| No | 18 languages | Not stated |
| Not stated |
Universal-2 (pre-recorded)AssemblyAI - Universal-2 | $0.15/hrassemblyai.com |
| No | 99 languages | Not stated |
| Not stated |
Universal-3.6 Pro RealtimeAssemblyAI - Universal-3.6 Pro Realtime | $0.45/hrassemblyai.com |
| Yes | 32 languages | Not stated |
| new |
Universal-Streaming (English)AssemblyAI - Universal-Streaming | $0.15/hrassemblyai.com |
| Yes | English only | Not stated | None recorded | Not stated |
Universal-Streaming MultilingualAssemblyAI - Universal-Streaming Multilingual | $0.15/hrassemblyai.com |
| Yes | 6 languages | Not stated | None recorded | Not stated |
Sync Speech-to-Text APIAssemblyAI - Sync API | $0.30/hrassemblyai.com |
| No | 18 languages | 2 minutes per request |
| new |
Dictation APIAssemblyAI - Dictation API | $0.62/hrassemblyai.com |
| Not stated | 19 languages | 120 seconds per request | None recorded | new |
Ink 2Cartesia - ink-2 | 3 credits per second of audio; $0.39 per hour on the Scale plancartesia.ai |
| Not stated | 5 languagesderived | Not stated |
| stable |
Whisper Large v3 (Groq)Groq - whisper-large-v3 | $0.111 per hourconsole.groq.com |
| No | Multilingual | 25 MB (free tier), 100MB (dev tier) |
| Not stated |
Whisper Large v3 Turbo (Groq)Groq - whisper-large-v3-turbo | $0.04 per hourconsole.groq.com |
| No | 99+ languages | 100 MB |
| Not stated |
| Voxtral Mini Transcribe 2Mistral | $0.003 /Min (cached $0.0003 /Min)docs.mistral.ai |
| No | 13 languages | up to 3 hours per request |
| Not stated |
| Voxtral Mini Transcribe RealtimeMistral | $0.006 /Mindocs.mistral.ai |
| Yes | 13 languages | Not stated |
| Not stated |
Qwen3-ASR-FlashAlibaba - qwen3-asr-flash | $0.000035/second (International)alibabacloud.com |
| Yes | Not stated | 10 MB and 5 minutes per file | None recorded | Not stated |
Qwen3-ASR-Flash-RealtimeAlibaba - qwen3-asr-flash-realtime | $0.000090/second (International)alibabacloud.com |
| Not stated | Not stated | Not stated | None recorded | Not stated |
Qwen-Audio-3.1-ASR-FlashAlibaba - qwen-audio-3.1-asr-flash | 0.15 USD/million input tokens; 0.47 USD/million output tokens (International)alibabacloud.com | Not convertedPriced per token | Not stated | Not stated | Not stated | None recorded | Not stated |
Fun-ASR (audio file recognition)Alibaba - fun-asr | $0.000035/second (International)alibabacloud.com |
| Not stated | Not stated | 12 hours and 2 GB per file (asynchronous) |
| Not stated |
FAQ
What is the cheapest speech-to-text API per hour?
Converting each speech-to-text price stated per second, minute or hour to dollars per hour (our arithmetic, not the providers'), the lowest on October 18, 2026 are Whisper Large v3 Turbo (Groq) at $0.04 an hour (stated as $0.04 per hour), Whisper Large v3 (Groq) at $0.111 an hour (stated as $0.111 per hour), Alibaba Fun-ASR (audio file recognition) at $0.126 an hour (stated as $0.000035 per second), Alibaba Qwen3-ASR-Flash at $0.126 an hour (stated as $0.000035 per second) and AssemblyAI Universal-2 (pre-recorded) at $0.15 an hour (stated as $0.15 per hour). Token-priced models (Gemini API and Alibaba) are left out because their cost depends on transcript length. Check minimums, regions and promotional rates on the linked pricing page.
How is the converted price per hour calculated?
A per-second price is multiplied by 3,600, a per-minute price by 60, and a per-hour price is shown as is. Regions, cached rates and other notes in parentheses are ignored, credit-only prices are skipped, and token-priced models are not converted. The provider's own figure is always shown next to it.
Which speech-to-text APIs support streaming?
18 of 39 models are documented as streaming, from OpenAI, Gemini API, Google Cloud, Amazon, Azure, ElevenLabs, Deepgram, AssemblyAI, Mistral and Alibaba. A "Not stated" means the pages we read do not say either way.
Verified from each provider's own pricing pages and model docs; every quote was string-matched against the page that day. Prices are shown in the provider's own unit. Values marked derived were counted by us or taken from Azure's Retail Prices API, not from a pricing page. Prices change without notice; check the linked source before you commit a budget to one.
Newsletter
Know when a speech API changes its prices
One email per week. New model releases, pricing changes and what they mean for what you build.
Keep exploring
Text-to-speech API pricing
Per character, per token and credit prices, grouped by unit.
Realtime voice APIs
Speech-to-speech models for live voice agents.
OpenAI deprecations
Shutdown dates for whisper-1 and the gpt-4o transcribe family.
Moving off whisper-1
What replaces the retiring OpenAI transcription models.