Skip to main content
Watch: Grok Bot Built My AI Video Arena While I Tested 8 Models

Speech APIs - verified

Speech AI APIs compared: speech-to-text, text-to-speech and realtime voice

We read the speech pricing pages and model docs of 12 providers on October 18, 2026 and recorded 79 models: 39 speech-to-text, 35 text-to-speech and 5 realtime voice. Every price stays in the provider's own unit, next to the line it was copied from. Converted to dollars per hour, the 35 speech-to-text prices stated per second, minute or hour run from $0.04 to $4.50. 5 models are scheduled to retire, the first on November 17, 2026.

Providers
12
Speech-to-text
39
Text-to-speech
35
Realtime voice
5

Compare by model type

Retiring soon

Shutdown dates from each provider's deprecation page, cross-checked against our deprecations tracker. Each badge links to the row.

Every provider

Each provider page lists every model with its verbatim quotes, sources and notes. Running a whole voice agent rather than picking a model? Compare voice AI agent platforms.

FAQ

What is the cheapest speech-to-text API?

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) and Alibaba Fun-ASR (audio file recognition) at $0.126 an hour (stated as $0.000035 per second). 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 text-to-speech API usage priced?

Per character: 27 models (Google Cloud, Amazon, Azure, ElevenLabs, Deepgram, Groq, Mistral and Alibaba). Per token: 7 models (OpenAI, Gemini API and Alibaba). Plan credits: 1 model (Cartesia). The units are not interchangeable, so this site lists them in separate tables and never converts one into another.

Which speech models are retiring soon?

November 17, 2026: Gemini API gemini-2.5-flash-preview-tts, Gemini API gemini-2.5-pro-preview-tts and Gemini API gemini-3.1-flash-tts-preview. January 6, 2027: OpenAI gpt-4o-mini-tts. February 26, 2027: OpenAI gpt-4o-mini-transcribe. Dates come from each provider's deprecation page, cross-checked against our deprecations tracker.

Which providers sell a realtime voice API?

OpenAI GPT-Realtime-2.1: $32.00 per 1M audio input tokens; $0.40 cached; $64.00 per 1M audio output tokens. OpenAI GPT-Realtime-2.1 Mini: $10.00 per 1M audio input tokens; $0.30 cached; $20.00 per 1M audio output tokens. OpenAI GPT-Live 1: $0.05 per minute. Gemini 3.8 Live: $3.00 per 1M audio input tokens ($0.005/min); $12.00 per 1M audio output tokens ($0.018/min). AssemblyAI Voice Agent API: $4.50/hr ($0.075/min). Each price is in the provider's own unit.

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.

Snapshot of October 18, 2026: 79 models, 35 speech-to-text prices converted to dollars per hour, 5 retiring.

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.