Tencent's open text-to-video model — cinematic clips you can run and fine-tune yourself.
Tencent's HunyuanVideo releases a full open text-to-video model with strong motion coherence and image-to-video support. Run it on your own GPUs, fine-tune with LoRAs, and avoid per-clip fees.
Who it's for: Studios and builders who want open, self-hostable, fine-tunable text-to-video without vendor lock-in.
The full model is open — run it on your GPUs or via hosted partners.
Strong temporal consistency and realistic motion versus earlier open models.
Seed it with an image and extend into a moving clip.
Train custom styles and characters with community LoRAs.
HunyuanVideo is the best open text-to-video model for teams that want to self-host and customize. Casual creators will prefer a hosted app like Kling or Sora, but for sovereignty and fine-tuning it leads the open field.