HunyuanVideo AI 视频生成器
Tencent's open-weight video generation family, important for developers, researchers, and self-hosted benchmarking.
是否适合使用 HunyuanVideo?
适合用来做： open-weight testing, developer experiments, self-hosted video generation, avatar and I2V research
Skip or compare first 的情况: License is not Apache-style; commercial usage must be checked. Consumer usability depends on wrappers and hardware. HF/模型Scope metrics are not collected yet.
定位： Tencent's open-weight video generation family for developers, researchers, and self-hosted benchmark users.
核心能力
- Open 13B baselineTencent released code and weights for a large open text-to-video model family.来源
- HunyuanVideo-1.5 lightweight focusThe 1.5 repository describes an 8.3B model designed to lower usage barriers.来源
- Distilled I2V speedupsThe 1.5 repository reports step-distilled I2V models that reduce generation time on RTX 4090-class hardware.来源
- Avatar ecosystemTencent maintains related HunyuanVideo avatar and I2V projects for human animation and reference workflows.来源
公司摘要
Built by Tencent
Founded / HQ: 1998 / China
What it does: Tencent is a global internet and technology company with products across social communication, games, digital content, cloud computing, fintech, and AI.
Main products/ecosystem: HunyuanVideo, HunyuanVideo-I2V, HunyuanVideo-Avatar, VideoCrafter
Founders/leadership: partial/evidence pending
Scale/funding/status: Status: Public 公司
Relationship to HunyuanVideo: Tencent Hunyuan publishes HunyuanVideo, HunyuanVideo-I2V, HunyuanVideo-Avatar, and HunyuanVideo-1.5. The company also has large gaming, social, cloud, and content ecosystems that make open-weight video strategically important.
价格、入口与开源状态
访问方式： Open 权重s, Closed Web, Aggregator
开源/闭源： Open.
Open weights are free to download under their license terms; practical cost depends on GPU, inference stack, and hosted wrapper.
Price, regional access, and commercial terms need a live check before cost-sensitive recommendations.
最新版本
HunyuanVideo-1.5
2025-11-21 / confirmed
Lightweight 8.3B video model line introduced. Supports text-to-video and image-to-video with lower usage barriers.
最新消息
Fp8 gemm inference is supported
2025-12-23 / Tencent-Hunyuan GitHub
External source link for this model.
打开原文关键得分驱动
These are the scoring dimensions most responsible for the current ranking position.
- 模型 Quality Proxy: 80HunyuanVideo is tracked by public leaderboards and has strong open-weight repository evidence, but current frontier placement needs live collection.
- Version Maturity: 995 version/history entries in the profile; latest public date is 2025-12-05.
- Capability Depth: 73Supported workflow tags: t2v, i2v, avatar_video, audio_driven_video, open_weights. Depth rewards T2V/I2V/V2V, references, native audio, editing, API, and open weights where present.
- Access And Pricing: 88Access types: open_weights, closed_web, aggregator. Pricing note: Open weights are free to download under the applicable license, but self-hosting cost depends on GPU memory and inference s...
替代选择与对比
Use these when price, access, output style, or workflow fit is uncertain.
- WanAlibaba's open and commercial video model family, with strong community adoption and variants for text, image, speech, and animation-driven video.View model对比
- LTX VideoLightricks' open-weight and productized video model family, notable for synchronized audio-video generation and production-oriented open deployment.View model对比
- CogVideoXAn influential open video generation model family from the THUDM/Zhipu ecosystem, useful as a research and self-hosted baseline.View model对比
- Runway GenA mature creative video generation platform and model family for production-style image-to-video, text-to-video, editing, and workflow automation.View model对比
它是什么
Primary content is rendered into static HTML.
HunyuanVideo is Tencent's open video generation model family. Its original 13B release became a major open baseline, and HunyuanVideo-1.5 later focused on lighter deployment and broader access for developers and creators.
For a consumer selection site, HunyuanVideo matters as an open-weight alternative to closed products like Kling, Seedance, Veo, and Runway. It can be run locally or through community hosts, but the consumer experience depends heavily on the UI, GPU setup, and wrapper used.
The family is also expanding beyond text-to-video into image-to-video, avatar, audio-driven human animation, and speed-distilled variants. That makes it useful for benchmark design, even when regular users do not directly self-host it.
版本进展
- HunyuanVideo-1.5Lightweight 8.3B video model line introduced. Supports text-to-video and image-to-video with lower usage barriers.
- HunyuanVideo 13BInference code and model weights released. Technical paper describes a large open-source video foundation model.
开源 / 来源证据
Status: Open
Tencent publishes code and weights; license is Tencent Hunyuan community license, not a permissive OSI-style license.
最新消息
模型-specific news entries link directly to the original publisher; article bodies stay off-site.
- Fp8 gemm inference is supported External source link for this model. 打开原文
- HunyuanVideo-1.5 release External source link for this model. 打开原文
- HunyuanVideo: A Systematic Framework For Large Video Generation 模型 Technical paper for the original open foundation model. 打开原文
完整得分拆解
权重ed evidence dimensions separate model quality from access, web authority, ecosystem, and source confidence. See methodology and rankings.
| 维度 | Score | 权重 | 证据 | Confidence |
|---|---|---|---|---|
| 模型 Quality Proxy | 80 | 24% | HunyuanVideo is tracked by public leaderboards and has strong open-weight repository evidence, but current frontier placement needs live collection. | high |
| Capability Depth | 73 | 16% | Supported workflow tags: t2v, i2v, avatar_video, audio_driven_video, open_weights. Depth rewards T2V/I2V/V2V, references, native audio, editing, API, and open weights where present. | high |
| Access And Pricing | 88 | 12% | Access types: open_weights, closed_web, aggregator. Pricing note: Open weights are free to download under the applicable license, but self-hosting cost depends on GPU memory and inference stack. Hosted platforms may cha... | high |
| Version Maturity | 99 | 12% | 5 version/history entries in the profile; latest public date is 2025-12-05. | high |
| Ecosystem Popularity | 100 | 10% | GitHub snapshot includes 2 repo(s), 16,946 stars, 1,537 forks, 216 open issues, latest pushed 2026-06-29T09:33:50Z. | high |
| Web Authority | 59 | 10% | SEO snapshot found product domain aivideo.hunyuan.tencent.com and company domain tencent.com; sampled sitemap URL count is 0. | medium |
| 公司 Distribution | 88 | 10% | Tencent distribution profile: HunyuanVideo is important for users and developers because Tencent released a strong open-weight video model line while also having large consumer, gaming, cloud, and content ecosystems.. | medium |
| 来源 Confidence | 100 | 6% | Profile source confidence is confirmed with 3 official URL(s), 4 feature evidence item(s), SEO present, GitHub present. | high |
限制与价格
Open weights are free to download under the applicable license, but self-hosting cost depends on GPU memory and inference stack. Hosted platforms may charge separately.