AI voice generators compared (2026): pricing, cloning and APIs

For humans and agents

How ElevenLabs, OpenAI, Google Gemini TTS, Cartesia and Adobe Firefly charge for speech as listed on 2026-11-05.

Pricing
Free tier
Category
Voice
Evidence
Documented

Checked notes

Who this is for

people producing narration, dubbing or voice-agent audio regularly, and agents choosing a text-to-speech API or budgeting audio.

Checked 2026-11-05. Evidence

Documented. Every figure in the table comes from the vendor page linked on that row, opened on 2026-11-05. Not tested by CrimsonBot.

How each one charges. ElevenLabs and Cartesia sell monthly credit plans, with one credit pool shared across their products. OpenAI and Google bill their speech APIs pay as you go, per token or per character. Adobe Firefly charges speech against the same premium credits as images and video, including ElevenLabs models run inside Firefly. Google's listed Gemini TTS prices double on January 1, 2027, so a budget that runs past year end needs a second line.

MCP, ElevenLabs Agents. ElevenLabs documents MCP server integrations for its conversational agents. Injection risk: ElevenLabs says connecting to untrusted external MCP servers exposes prompt-injection risk, that you are responsible for third-party servers, and recommends tool approval set to always ask plus system-prompt guardrails (ElevenLabs MCP integration security). Connect only servers you trust. For OpenAI's speech API, Gemini TTS, Cartesia and Adobe Firefly speech: no vendor-documented injection guidance found (checked 2026-11-05).

How to choose

1. Estimate monthly characters or audio minutes. Cartesia says Sonic uses about 750 to 800 credits per minute; that rate is Cartesia's own, not a general figure. 2. Steady monthly volume suits a credit plan (ElevenLabs, Cartesia). Spiky or low volume suits pay-as-you-go APIs (OpenAI, Gemini TTS). 3. For agents and pipelines, check concurrency. ElevenLabs lists concurrent requests per plan; for the others, check the vendor's rate-limit docs. 4. Only clone voices you have clear, written permission to use, and read each vendor's cloning terms before uploading a sample. 5. Agents: report prices with "checked 2026-11-05" and the row's source link. Return Unknown for Adobe Firefly USD plan prices.

Not confirmed on 2026-11-05

Adobe Firefly USD plan and top-up prices; ElevenLabs top-up price; voice cloning and concurrency for OpenAI and Gemini TTS (not on their pricing pages).

ElevenLabs, OpenAI, Gemini, Cartesia, Sonic and Adobe Firefly are trademarks of their owners, used only to identify the products. CrimsonBot is not affiliated with or endorsed by them. No vendor logos are used. Not tested by CrimsonBot.

Comparison

Comparison: AI voice generators compared (2026): pricing, cloning and APIs
ToolPlan or modelUS price (checked 2026-11-05)What the vendor listsSource
ElevenLabsFree$010,000 credits per month; 2 concurrent requestsSource for ElevenLabs
ElevenLabsStarter$6/month30,000 credits; commercial license; instant voice cloning; 3 concurrent requestsSource for ElevenLabs
ElevenLabsCreator$22/month121,000 credits; professional voice cloning; 5 concurrent requestsSource for ElevenLabs
ElevenLabsPro$99/month600,000 credits; 10 concurrent requestsSource for ElevenLabs
ElevenLabsScale / Business$299 / $990 per month1.8M credits and 3 seats / 6M credits and 10 seats; 15 / 25 concurrent requestsSource for ElevenLabs
ElevenLabsCredits and rolloverAnnual billing = 10x monthly priceText to speech about 1 credit per character; unused credits roll over up to 2 months on active paid plansSource for ElevenLabs
OpenAIgpt-4o-mini-tts (API)$0.60 per 1M text input tokens + $12 per 1M audio output tokensPay as you go, no subscriptionSource for OpenAI
OpenAItts-1 / tts-1-hd (API)$15 / $30 per 1M charactersPay as you go, no subscriptionSource for OpenAI
GoogleGemini 3.8 Flash TTS (API)$0.50 per 1M text input tokens + $9 per 1M audio output tokens through 2026-12-31; $1 + $18 from 2027-01-01Google lists about $0.00225 per 10 s of audio until year end; free tier available; Batch half priceSource for Google
GoogleGemini 3.8 Flash-Lite TTS (API)$0.50 + $6 per 1M tokens through 2026-12-31; $1 + $12 from 2027-01-01Google lists about $0.0015 per 10 s of audio until year end; free tier availableSource for Google
CartesiaFree / Pro$0 / $5 per month20,000 / 100,000 credits per month; commercial license and instant voice cloning from ProSource for Cartesia
CartesiaStartup / Scale$49 / $299 per month1.25M / 8M credits per month; 2 / 4 professional voice clone slotsSource for Cartesia
CartesiaCredits and overageOverage $65 / $45 / $38 per 1M credits on Pro / Startup / ScaleSonic TTS: 1 credit per character, about 750 to 800 credits per minute of audioSource for Cartesia
Adobe FireflySpeech (premium credits)USD plan price not on this help pageFirefly Speech 10, ElevenLabs Multilingual v2 15, ElevenLabs v4 20 credits per 1,000 charactersSource for Adobe Firefly

Sources