Video generation

HunyuanVideo (13B)

Name: HunyuanVideo (13B)
Author: Tencent Hunyuan

Tencent's original open video foundation model, with over 13 billion parameters - the largest open-weights video model at its December 2024 release.

A video generation model by Tencent Hunyuan · read 2026-08-14 · sourcing high

Resolution: 720x1280
Aspect ratios: 9:16, 16:9, 4:3, 3:4
Weights: Downloadable
API: Yes
Commercial use: Permitted subject to conditions

Generates 129-frame clips at 540p to 720p across 9:16, 16:9, 4:3, 3:4 and 1:1 aspect ratios. Now largely superseded by HunyuanVideo 1.5 but remains the base for many derivative models.

Commercial use — Permitted subject to conditions

Tencent Hunyuan Community License Agreement. Commercial use is permitted outside the EU, UK and South Korea.

Not licensed in: European Union, United Kingdom, South Korea

The licence does not apply in some territories. The licence does not apply in the European Union, United Kingdom or South Korea.
Permission required above a user threshold. Requires Tencent's explicit permission above 100 million monthly active users.
Output may not be used to train AI. Forbids using the works to improve any other AI model.

This entry was written by hand from the licence wording quoted above, because the source dataset's yes/no field was misleading here. Read the licence before you rely on any of this. Nothing on this page is legal advice, and this site does not tell you a model is cleared for commercial use.

Specification

Vendor figures, read 2026-08-14Read from Tencent Hunyuan's own documentation on 2026-08-14. Vendors change these between model versions without announcing it, so check theirs before you commit a shot to a number. Sourcing for this entry is strong: the figures came from pages we retrieved and read. Check the vendor's current documentation

What it takes in: text
What it makes: video
Maximum duration: not established
Maximum resolution: 720x1280
Aspect ratios: 9:16, 16:9, 4:3, 3:4, 1:1
Controls: text prompt, negative prompt, resolution and aspect ratio presets, flow-shift parameter, inference steps, seed, LoRA fine-tuning, xDiT multi-GPU parallel inference
Weights: Downloadable
API: Yes
Pricing: Free open weights under the Tencent Hunyuan Community License. Self-hosting requires about 60GB peak GPU memory at 720x1280 and about 45GB at 544x960 per the official README, so an A100 80GB or H100 class GPU; community quantizations and offloading reduce this. Hosted access via Tencent Cloud.
Released or updated: 2024-12-03
Prompt limit: 512 tokens (T5-XXL text encoder)Vendor figure, checked 2026-08-16Collected from the vendor's own documentation on 16 August 2026. Vendors change these between model versions without announcing it, so check the vendor's current API reference before you rely on it. Check the vendor's documentation

Maximum duration is not established here. That means no page we could retrieve stated it, not that the model lacks one.

Strengths

Documented aspect-ratio support across five ratios
Large ecosystem of fine-tunes, LoRAs and derivative models built on it
Training-adjacent tooling and multi-GPU inference officially supported
Tencent waives rights to outputs

Weaknesses

EU/UK/South Korea territorial exclusion
60GB VRAM at 720p
No fps stated in official docs
Superseded by HunyuanVideo 1.5 on both quality and hardware cost
No audio

Best for

Derivative model development and fine-tuning where the 13B backbone's ecosystem matters more than raw quality.

Where the sources disagree

Recorded rather than resolved. Picking the tidier number would hide the fact that the vendor's own pages do not agree.

fps is not stated on the GitHub README or Hugging Face card; secondary sources commonly say 24 fps, which would make 129 frames about 5.4 seconds. Left null.

What to understand before using it

Explainers naming the mechanisms this model's own documentation mentions.

LoRAs and fine-tuning: teaching a model your subject names “fine-tuning”
Training data and copyright: what is not settled names “training”
Open weights versus closed API: renting or owning the model names “weights”

MCP servers that front Tencent Hunyuan

Servers in the media MCP directory whose own description or homepage names this vendor. Absence of a server here is not evidence that none exists: most publish no tool list at all.

3daistudio 3D and animation
MCP server for 3D AI Studio, generate 3D models from text or images using Hunyuan and TRELLIS.
Corent Image generation
Generate images and videos via the Corent API, auto model routing with provider fallback.
PicoBerry 3D and animation
Generate game-ready 3D models, images, and animations. Multi-engine, one API.
Picsart GenAI Image generation
Generate and edit images, videos and audio with 150+ AI models.
Replicate Image generation
Replicate MCP server - image, video, audio, speech, LLM, vision, upscale, 3D, training. 36 tools.
Shortgenius Video generation
Generate AI short-form videos and auto-publish them to TikTok, YouTube, Instagram, Facebook and X.

See it from the vendor

Shotcraft hosts no sample output. Judging a model means watching its own showreel with your own eyes, and a clip re-hosted here would be a clip we chose.

Nearby in the directory

HunyuanVideo 1.5

An 8.3B-parameter video model covering both text-to-video and image-to-video, released with 480p and 720p checkpoints plus a super-resolution stage th

Video generation

HunyuanVideo-Avatar

Audio-driven human animation built on HunyuanVideo. Given a character image and an audio track it generates emotion-controllable dialogue video, and s

Video generation

HunyuanVideo-I2V

An image-to-video adaptation of HunyuanVideo that uses a token-replace technique to condition on a reference image, emphasizing first-frame consistenc

Video generation

Sources

Read on 2026-08-14. A source that has changed since then will not be reflected here until the next pass.