ElevenLabs Chimera: what it is and who can use it

ElevenLabs Chimera is a new feature in ElevenCreative, the company's audio and video workspace, that makes a character talk on camera. You give it a face and a line, and it gives you back a video where the voice and the expression match. It runs on two models, ElevenLabs' own Eleven v4 for the voice and ByteDance's Seedance 2.5 for the video, and that second one is why a lot of people reading this in the US probably can't use it yet.

The launch post

@ElevenCreative, the account for ElevenLabs' creative tools, announced Chimera on October 21 with a 50-second 4K demo reel. The post opens with "What if your character could finally say it themselves?" and then explains who does what: Eleven v4 puts the emotion in the voice, and Seedance 2.5 carries it onto the face. When we checked on October 22 it had about 155,000 views and nearly as many bookmarks as likes, which usually means people are saving it to try later.

Creators were doing this before Chimera existed. Three weeks earlier, @jerrod_lew, an Australia-based artist who teaches AI video, posted a demo of Eleven v4 and Seedance 2.5 used together and called it "an amazing combo." He had to stitch the two together himself. Chimera does that in one step inside ElevenLabs.

We wrote about the voice half when it came out: Eleven v4 took first place in blind listener tests, and it costs a good deal more than Google's and Cartesia's voices.

What Chimera can do

The launch post lays out a short workflow:

1. Open ElevenCreative and create or upload a character.
2. Write the line you want them to say.
3. Pick a voice from the library, or design a new one.
4. Add audio tags such as [whispers], [laughs] or [yells] to control timing and mood word by word.
5. Generate. The face acts out the read.

ElevenLabs' Seedance 2.5 page says the video model makes 30-second narrative clips in one pass, takes up to 50 reference images, videos or audio files, and can edit one region of a frame without redoing the whole clip. ElevenLabs hasn't published a Chimera page, a spec sheet or a per-video price. Anything past what the launch post says is unconfirmed.

The US catch

That same Seedance 2.5 page says the model isn't available to ElevenLabs users in the United States, and it blames its China-based provider. ElevenLabs' image and video docs mark most ByteDance, Kling, MiniMax and Wan video models as unavailable in the US too.

Chimera depends on Seedance 2.5, and ElevenLabs hasn't announced an exception for it. If you run a business in Illinois, plan on not seeing the feature until the company says otherwise. Creators in other countries can use it today.

How to get a similar result in the US

You can still make a talking-character video inside ElevenLabs. It's two steps instead of one, with a different video model. Write and voice the line in Eleven v4 with its audio tags, which work in the US, then give that audio to one of the models below. None of them is on the US-restricted list in ElevenLabs' docs.

You need a paid plan for video. ElevenLabs' pricing page lists Starter, at $6 a month for 30,000 credits, as the cheapest plan that includes it. Credit cost per video depends on the model, its settings and the clip length, so make one short test clip before you plan a batch.

My read

Chimera doesn't add a new model. Both models already existed, and creators were already pairing them. What ElevenLabs is selling is one workspace and one bill, which is worth a lot if you make short clips every week. It also means the US restriction takes away most of what's new here: Americans can still get the voice, just without the shortcut.

For a small business, a talking character mostly makes sense for clips you'd otherwise record over and over: a 20-second "how to prep for your appointment" video, a product walkthrough, the same message in Spanish and Polish. Customers notice the voice first, and that part works in the US. Be careful with faces. If you animate a real employee, get written consent first. Our post on voice cloning and Illinois law explains why this state is stricter than most.

What it signals for businesses

AI video keeps getting cheaper and quicker to make, while keeping track of which model works in which country keeps getting harder. Before you build a routine around a feature, check what model sits underneath it and where that model is allowed to run. If you're also weighing AI for the phone, our look at Gemini Live Avatar and Qwen's 95 percent price cut covers the cheap end of the voice market.

If you're wondering where short video or voice could save your team time, New Face Design's free process audit is a good place to start. We'll go through your week with you and point out the few tasks worth automating.