Hunyuan Video V1.5 Text to Video

fal-ai/hunyuan-video-v1.5/text-to-video
Hunyuan Video 1.5 is Tencent's latest and best video model
Inference
Commercial use

Input

Additional Settings

Customize your input with more control.

Result

Idle
This generation takes approximately 3m.

What would you like to do next?

Current pricing is 0.075 cents/s of video, more resolutions arriving soon.

Logs

Hunyuan Video 1.5 | [text-to-video]

Tencent's Hunyuan Video 1.5 delivers up to 121 frames of 480p video at $0.075 per second of output. Trading resolution for prompt adherence and motion quality, the model generates approximately 5 seconds of video in 3 minutes. Best suited for rapid prototyping, social media content, and iterative creative workflows where speed and cost efficiency matter more than 4K output.

Use Cases: Social Media Content | Concept Visualization | Rapid Prototyping


Performance

At $0.075 per second of video output, Hunyuan Video 1.5 positions as a cost-effective text-to-video option for standard-definition workflows, with pricing roughly 13 generations per $1.00 on fal.

MetricResultContext
Resolution480p (16:9 or 9:16)Standard definition optimized for web/social
Inference Speed~3 minutesFor 121-frame generation
Cost per Second$0.07513.3 seconds per $1.00 on fal
Max Duration121 frames (~5 seconds)At default frame count
Related EndpointsHunyuan VideoStandard variant of same model family

Built for Prompt Precision Over Resolution

Hunyuan Video 1.5 uses Tencent's diffusion architecture optimized for semantic understanding rather than pixel density. Unlike models that prioritize 4K output at the expense of prompt adherence, this approach focuses computational resources on motion coherence and text interpretation.

What this means for you:

  • Prompt Expansion Built-In: Optional automatic prompt enhancement improves scene detail and motion quality without manual prompt engineering

  • Flexible Aspect Ratios: Native 16:9 and 9:16 support eliminates cropping for vertical social content or horizontal web video

  • Reproducible Generations: Seed control enables exact recreation of successful outputs for iterative refinement workflows

  • Controlled Inference Steps: Adjustable from 1-50 steps (default 28) lets you trade generation time for quality based on project requirements


Technical Specifications

SpecDetails
ArchitectureHunyuan Video 1.5
Input FormatsText prompts with optional negative prompts
Output FormatsMP4 video file
Resolution Options480p (16:9 or 9:16 aspect ratio)
LicenseCommercial use permitted

API Documentation | Quickstart Guide | Enterprise Pricing


How It Stacks Up

Hunyuan Video – Hunyuan Video 1.5 shares the same base architecture with updated inference optimizations. The standard Hunyuan Video variant offers alternative resolution/duration configurations for different project requirements at comparable pricing.

LongCat Video 720p – Hunyuan Video 1.5 prioritizes cost efficiency at 480p resolution for high-volume workflows. LongCat trades higher per-second costs for 720p output when resolution requirements justify the premium.

Bytedance SeeDance Pro Fast – Hunyuan Video 1.5 emphasizes prompt expansion and semantic understanding through Tencent's architecture. SeeDance Pro Fast focuses on speed optimization with different inference characteristics for time-sensitive generation needs.