Guides / SoulX-FlashHead 1.3B (real-time talking heads)

model

SoulX-FlashHead 1.3B (real-time talking heads)

AI-generated illustration, not actual output from this listing

What they did

A 1.3B streaming talking-head model. The Lite version reaches 96 FPS on one RTX 4090 (or three real-time streams), and the Pro version runs at 10.8 FPS on a 4090 or real-time on two RTX 5090s. Inference code and weights are open.

Worth knowing

  • Price: Open weights on Hugging Face (other)
  • License: other
  • Status: sourced. We have not run this ourselves yet (no "Tested" badge).
  • Quality bar (computed by code): 4/5 (primary source ✓, reproducible ✗, new ✓, example output ✓, safe ✓)
Get Atlas Dispatch

The best new AI video workflows and prompts, weekly.

Weekly: 5 new prompts and workflows. Unsubscribe anytime.
For your agent

Give your AI agent a video director

Connect Claude, ChatGPT, Cursor, Gemini or Grok Bot to Atlas, and your agent can plan shots, pick models and price them from the data on this site. Free. No sign-up or API key.

Step 1 of 4

Which tool do you use?