ElevenLabs brings image, video, and audio tools together for creators

In 2026, ElevenLabs expanded its creative platform beyond voice. Its September announcements include image, video, and template APIs, multimodal generation through MCP, and a new text-to-speech generation, Eleven v4. For creators building an AI character or influencer, the main change is being able to connect more production steps. The right tool still depends on the output you need.

What arrived at ElevenLabs in September 2026?

On September 26, ElevenLabs announced image, video, and template APIs, updated on October 12. According to its official API announcement, the platform provides access to models from Google, OpenAI, ByteDance, and Creatify for generating and editing images, animating them into video, and running templates. This describes a catalog available through ElevenLabs; it does not mean every model was built by ElevenLabs.

Templates are reusable workflows built in Flows: a team sets inputs, brand references, models, and output formats, then reruns the process with new data. The published example is ecommerce: provide a product image and name to get catalog photos, ad variants, and a product video. It is a workflow for teams that repeat production at scale, not a promise that every character will keep the same face in every generation.

On September 14, the company also announced voice, music, sound effects, image, and video generation through the ElevenLabs MCP. The connector lets people request these tasks in a conversation and sends the output to an ElevenCreative workspace. The company says the MCP brings together more than 50 models; it is a platform connecting them, not one proprietary multimodal model.

What about Eleven v4?

The September 28 announcement, updated October 19, introduces Eleven v4 and v4 Turbo as text-to-speech models. The company highlights emotion and context interpretation, voice consistency, and support for more than 90 languages. Those capabilities matter for narration, dialogue, audiobooks, and voice agents. They do not mean v4 is the model that generates video.

That distinction helps when choosing a workflow. If you need repeatable product-photo variations, evaluate image templates. For a piece with narration, music, and generated scenes, MCP or Studio can bring steps together. If your goal is a character who looks like the same person across posts, compare visual consistency across outputs and test different outfits, framing, and expressions with approved references. Voice consistency and face consistency are related goals, but different components handle them.

How is this different from Duble?

Duble has a narrower task: choose a character and put it in place of someone in a trend or a video that has already been recorded. For example, the original Passinho do Jamal clip and the version with Vector show the character inside the same scene. The generated video comes with sound; review speech, singing, and music in the result.

So ElevenLabs and Duble are not two interfaces for the same output. ElevenLabs announces general creative tools, audio, APIs, and reusable workflows; Duble is a character replacement workflow for real video. No ElevenLabs integration with Duble is announced. To compare other categories, read talking avatar or character in real video and AI influencer generators.

Before choosing, define the final asset: a catalog image, a video generated from text, narration, or a character inserted into an existing recording. Then compare identity control, editing, formats, cost, and how much of the process you need to automate. To start with Duble and a real video, open the app and choose a character.