ElevenLabs API Guide: Pricing, Voice Cloning & Code (2026)

ElevenLabs has become one of the go-to platforms for AI voice: text to speech that sounds genuinely human, voice cloning that takes seconds to set up, and a full conversational AI toolkit for building voice agents. If you are a developer trying to figure out what the ElevenLabs API actually offers, what it costs, and how to make your first request, this guide covers all of it, including a working code example. We will also touch on why voice generation pairs so naturally with AI video and avatar pipelines, since that combination is where a lot of interesting products are being built right now.

What Is the ElevenLabs API

The ElevenLabs API gives developers programmatic access to the same voice AI engine that powers the ElevenLabs web app. That includes text to speech, speech to speech (voice conversion), voice cloning, voice design, and a conversational AI layer for building voice agents that can hold real phone or chat conversations.

It is built for developers who want to add narration, dubbing, voice assistants, or synthetic voiceovers to their own products, rather than using ElevenLabs as a standalone tool. You authenticate with an API key, send a request with your text and a chosen voice, and get audio back. From there you can build anything from an audiobook pipeline to a customer support voice bot. If you need voice to be just one part of a larger product that also handles images or video, a unified AI API makes it much easier to manage everything in one place.

Key Features

Text-to-speech and multilingual voices

ElevenLabs offers a few different models depending on what you need. Eleven v3 is the most expressive and supports over 70 languages, though it caps out at 5,000 characters per request. Multilingual v2 covers 29 languages and is the more stable choice for longer form content, supporting up to 10,000 characters. Flash v2.5 trades a bit of expressiveness for speed, supporting 32 languages with roughly 75 millisecond latency, and it costs about half as much per character since it uses a lighter credit rate.

The voice library itself is large, with thousands of community shared voices you can browse and use directly, on top of whatever custom voices you create.

Voice cloning and voice design

This is the feature most people come to ElevenLabs for. Instant Voice Cloning lets you generate a usable clone from a short audio sample, and it is available starting on the Starter plan. Professional Voice Cloning produces a higher fidelity clone and unlocks on the Creator plan and above. There is also Voice Design, which lets you generate an entirely new, non-cloned voice just from a text description, useful when you want a distinct voice without recording anyone.

Real-time and streaming

For voice agents and live applications, latency matters. ElevenLabs offers streaming and WebSocket endpoints, and the Flash models in particular are tuned for near real-time responses. This is the backbone of ElevenLabs' Conversational AI product, which handles full voice agents, phone calling through SIP trunk or Twilio, and knowledge base integration for building support or sales bots that actually sound natural.

ElevenLabs API Pricing

ElevenLabs uses a monthly credit system, and how far your credits go depends on which model you use. On the higher quality Multilingual models, one character costs one credit. Flash models cost roughly half a credit per character, so the same credit balance stretches about twice as far if you are willing to trade a little expressiveness for speed and cost.

Here is the current tier breakdown:

The free tier does not include commercial usage rights, so anything you plan to ship needs at least the Starter plan. If you go over your monthly credits, pay-as-you-go rates kick in, for example around $0.05 per 1,000 characters on Flash and $0.10 per 1,000 characters on Multilingual v2. Annual billing knocks off roughly two months' worth of cost compared to paying monthly.

How to Get API Access

1. Create a free account at elevenlabs.io. No card is required to get started on the free tier.
2. Go to your account settings and generate an API key. You can scope keys to specific permissions and even set credit limits or IP restrictions on them, which is worth doing before you put a key into production.
3. Every request needs the header xi-api-key set to your key.
4. Pick a voice ID from the voice library (or one you have cloned), pick a model, and you are ready to make your first call.

Code Example: Generating Speech from Text

Here is a minimal example using the official Python SDK to convert text to speech and save it as an MP3.

import os
from dotenv import load_dotenv
from elevenlabs.client import ElevenLabs

load_dotenv()

elevenlabs = ElevenLabs(api_key=os.getenv("ELEVENLABS_API_KEY"))

audio = elevenlabs.text_to_speech.convert(
    text="The first move is what sets everything in motion.",
    voice_id="JBFqnCBsd6RMkjVDRZzb",  # "George" from the voice library
    model_id="eleven_multilingual_v2",
    output_format="mp3_44100_128",
)

with open("speech.mp3", "wb") as f:
    for chunk in audio:
        f.write(chunk)

Install the SDK first with pip install elevenlabs python-dotenv. If you would rather skip the SDK, the same request works as a plain cURL call to https://api.elevenlabs.io/v1/text-to-speech/{voice_id} with your API key in the xi-api-key header and a JSON body containing text and model_id.

ElevenLabs API Alternatives

OpenAI's TTS models are the most common comparison point. They are noticeably cheaper per character and simple to use, but they do not offer voice cloning and generally sound less emotionally expressive than ElevenLabs, especially on longer passages. Other options like Google Cloud TTS, Amazon Polly, and newer entrants like Cartesia tend to be faster and cheaper still, but again without the cloning quality that makes ElevenLabs stand out. If your project involves AI music generation alongside voice — think background music, jingles, or song creation — check out our guides on the Suno API, Udio API, and Mureka API.

A common pattern worth mentioning: voice is rarely the whole product. Teams building AI avatars, dubbed video, or talking head content usually need to pair a voice API with an image or video model for lip sync and visuals. Rather than integrating each vendor separately, a unified API like Apiframe gives you one API for the image and video side (Midjourney, Kling, Runway, and others), which cuts down on the integration work when voice is just one piece of a larger pipeline.