Hunyuan Video 混元视频
Tencent's open-weights video model, on consumer GPUs since version 1.5
Overview
Hunyuan Video is Tencent's open-weights text-to-video and image-to-video family. The current HunyuanVideo-1.5 (November 2025, 8.3B parameters) runs on consumer GPUs and ships under the Tencent Hunyuan Community License, which excludes the EU, UK and South Korea; Tencent Cloud also sells hosted generation by points.\n\n---\n\n混元视频是腾讯的开源权重视频生成模型系列。当前版本 HunyuanVideo-1.5（2025年11月，83亿参数）可在消费级显卡运行，采用腾讯混元社区许可协议，该协议不适用于欧盟、英国和韩国；腾讯云另提供按积分计费的托管服务。
Pricing
Pricing shown for reference only. These figures reflect RECATOOLS research as of 14 Oct 2026 and may be out of date or incomplete. This is not financial or purchasing advice — always confirm the current price on the provider’s official website before making any decision.
- HunyuanVideo-1.5 text- and image-to-video weights
- Training and LoRA scripts
- Not licensed in the EU, UK or South Korea
- 5s 720p clip: 1.5 points
- 5s 1080p clip: 3 points
- Failed tasks not charged
- Must be switched on in the console
- Concurrency add-ons sold separately
Use cases
What you can produce with Hunyuan Video 混元视频
- Text-to-video clips from an open 8.3B model, upscaled to 1080p
- Image-to-video animation that keeps the source image's look
- Faster 480p image-to-video with the step-distilled model
- LoRA or full fine-tunes from the published training code
- Local generation on a single 14 GB consumer GPU with offloading
- Hosted clips through Tencent Cloud without running GPUs
ASEAN Perspective
Hunyuan Video 混元视频 in Southeast Asia
开源 video 模型推动国产视频生态形成下游应用层——许多视频生成 App 直接基于其继续微调。
HunyuanVideo-1.5 is one of the most practical open video models to self-host: 8.3 billion parameters, 14 GB of GPU memory with offloading, a step-distilled fast mode, training scripts and Diffusers support. For fine-tuning and research it is a strong base.
Two cautions. The licence excludes the EU, UK and South Korea outright and gates the very largest companies, so check where your users are before building on it. And the launch quality claims came from Tencent's own evaluation of an unreleased version. No new video weights have shipped in 2026.
What we found
Hunyuan Video is Tencent's open-weights video generation family. The original HunyuanVideo arrived on 3 December 2024 with more than 13 billion parameters, which Tencent described as the largest open-source video model at the time, and a human evaluation over 1,533 prompts in which it beat Runway Gen-3 and Luma 1.6. Read that claim carefully: Tencent's own README says the evaluation used a high-quality version different from the one it released, and the released model needed about 60 GB of GPU memory for 720p.
HunyuanVideo-1.5, released on 20 November 2025, is the current model. At 8.3 billion parameters it runs on a consumer card with 14 GB of memory when offloading is enabled, generates from text or an image, and upscales to 1080p. In December 2025 Tencent added a step-distilled 480p image-to-video model that it says cuts generation time by 75% on an RTX 4090, published the training and LoRA scripts, and added FP8 inference and Diffusers support. Companion repositories cover image-to-video, talking avatars and Foley sound. No new video weights have appeared in 2026; Tencent's Hy AI Studio, where the old hunyuan.tencent.com address now lands, still lists HY Video 1.5.
The licence is the part to read first, because it is not open source in the usual sense. The Tencent Hunyuan Community License does not apply in the European Union, the United Kingdom or South Korea, so use there is unlicensed, and any company whose products had more than 100 million monthly users at release must ask Tencent for a licence.
For those who do not want to run GPUs, Tencent Cloud sells a hosted 混元生视频 service by points: a five-second 720p clip costs 1.5 points and 1080p three, with prepaid packs from ¥5,000 for 5,000 points or pay-as-you-go at ¥1.2 a point. Tencent says those services are moving to its TokenHub platform. For fine-tuning research and self-hosted generation outside the excluded territories, it remains one of the strongest open video models.
Compiled by RECATOOLS from public sources: vendor documentation and pricing, code repositories and published reporting.
Frequently asked questions
About this listing
This entry was compiled from publicly available data including Hunyuan Video 混元视频's official website, press releases, documentation, and reputable third-party publications. RECATOOLS is not affiliated with Hunyuan Video 混元视频 unless explicitly stated.
Third-party AI tools update their pricing, features, availability, and policies frequently. Information here may be outdated by the time you read this — we make reasonable efforts to keep listings current, but cannot guarantee absolute accuracy.
For the latest details, please refer to Hunyuan Video 混元视频 directly →
Spotted something out of date? Suggest an update →
Alternatives to Hunyuan Video 混元视频
More in Video & Audio
