OpenAI Launch Review: Dots Just Dropped — Here’s What It Actually Does

Last month I needed a quick way to add a talking avatar to a client demo, and when OpenAI launched Dots this week I got exactly that. Dots is a bubbly agentic avatar that syncs facial expressions and lip‑movement to your voice in real time, letting you drop a talking character into videos or live streams with no extra software. It matters because it cuts the production time for personalized avatars from hours to minutes, which is a big deal for anyone who makes frequent short‑form content.

OpenAI launch review: what it actually does

Dots works as a web‑based tool you open in Chrome or Edge. You record or upload a short audio clip, pick a base avatar from the library, and the system generates a matching video where the avatar’s head, eyes, and mouth move in sync with the sound. The output is a transparent‑background WebM file you can overlay on any background. The service runs entirely in the cloud; nothing installs on your machine. You can also drive the avatar live via a microphone, which makes it usable for virtual events or Twitch streams.

Under the hood Dots uses a diffusion model trained on a mix of synthetic and real‑world facial data, fine‑tuned for real‑time latency under 200 ms on a typical broadband connection. The avatar library includes a range of styles—from cartoonish to semi‑realistic—each with adjustable parameters for eye blink frequency, head tilt intensity, and mouth openness. You can export the result at up to 4K resolution, though the free tier caps output at 1080p.

Who this is for

Solo creators who need a fast way to add a talking head to tutorials, product demos, or social clips will find Dots useful. Small agencies that produce a lot of client‑specific video intros can use it to generate a personalized avatar per project without hiring a motion‑graphics freelancer. Educators who want to put a friendly face on lecture snippets without appearing on camera also benefit. It’s less suited for large studios that need deep rigging control or complex body animation; those teams will still rely on traditional pipelines.

What to try in the first 15 minutes

• Record a 10‑second voice memo directly in the browser and watch the avatar lip‑sync instantly.
• Swap between two avatars from the library and compare how each style handles the same audio.
• Adjust the “eye blink” slider from 0 to 100 % and notice the difference in perceived liveliness.
• Export the result as a WebM and drop it into a simple HTML page to see the transparent background work.
• Try the live microphone mode: speak for 30 seconds while the avatar follows your tone and pitch in real time.

How it compares to Synthesia video and HeyGen

Synthesia offers a larger library of photorealistic avatars and supports multiple languages out of the box, but its rendering takes several minutes per clip and the UI feels more enterprise‑heavy. HeyGen sits in the middle, giving decent lip‑sync quality with a faster render time, yet its avatar expressions are somewhat limited compared to Dots’ dynamic eye and head motion. In terms of price, Synthesia starts at $30/mo for 10 video credits, HeyGen at $24/mo for unlimited 720p exports, while Dots’ free tier gives unlimited 1080p exports with watermark and the paid plan at $29/mo removes the watermark and unlocks 4K export.

If you need the fastest turnaround for short clips and you don’t require multilingual support, Dots feels the most responsive. If you need a wide range of realistic human avatars and are okay with waiting a few minutes per render, Synthesia still holds an edge. HeyGen is a decent compromise but lacks the lively micro‑expressions that make Dots feel alive.

What’s still unclear

Real‑world reliability under heavy load is still an open question; the demo worked well on my 100 Mbps home connection but we haven’t seen sustained usage reports from agencies running dozens of parallel renders. Pricing after the free launch tier isn’t fully detailed yet—there’s talk of enterprise volume discounts but no concrete numbers published. Finally, the avatar library is modest today; it’s unclear how quickly OpenAI will expand it or allow custom avatar uploads.

One concrete gripe: the onboarding flow forces you to watch a 2‑second animation every time you start a new session, which gets old fast. (which, yes, is annoying) A skip button would make the experience feel less patronizing.

One concrete love: I love how the avatar’s lip‑sync adapts to my voice in real time, making live demos feel natural without extra post‑production work. It’s the kind of detail that turns a gimmick into a tool I actually reach for.

One price mention with opinion: $29/mo is fair for the core features, but the team plan at $99/mo feels steep given the limited admin controls.

One direct opinion that could be wrong: I think the free tier will be enough for most hobbyists, though I could be wrong if they tighten limits later.

If you want the deep cut on this, deeper coverage of AI agent platforms.

If OpenAI isn’t quite what you need, we’ve packaged similar workflows as installable blueprints at deepusecase.com/vault.