Fast, reliable, and cost-efficient.
fal adapts to your usage, ensuring you only pay for the computing power you consume. It's cost-effective scalability at its best.
GPU Pricing
Deploy your own app on our competitively priced GPU fleet.
GPU Pricing
Competitive pricing for custom deployments. Get H100s from as low as $1.99/hr. Contact support@fal.ai to get started.
| GPU | VRAM | Price per hour* | Price per second* | |
|---|---|---|---|---|
 H100 | 80GB | $1.89/h | $0.0005/s | |
 H200 | 141GB | $2.10/h | $0.0006/s | |
 A100 | 40GB | $0.99/h | $0.0003/s | |
 B200 | 184GB | contact us | contact us |
*starting at
Output-Based Pricing
Models deployed by us are priced by the output they generate, like videos, images or tokens
Note: Some models not shown here may use GPU-based pricing even if they generate media.
Video Models
Video models are billed by output unit — per second or per video — depending on the model. The values shown represent output-based pricing.
| Model | Unit | Price | Output per $1 | |
|---|---|---|---|---|
| Hunyuan Video | video | $0.4 | 3 videos | |
| Kling 1.6 Pro Video | video second | $0.095 | 11 video seconds | |
| Kling 2 Master Video | video second | $0.28 | 4 video seconds | |
| Alibaba Wan Video | video | $0.4 | 3 videos | |
| MiniMax Video Live | video | $0.5 | 2 videos |
* For a fair comparison, we've normalized these values to show approximate output per $1.
* Based on an estimated average video = 5 seconds at 720p. Actual output may vary based on model, resolution, and prompt complexity.
Image Models
Image models are billed by either image count or output size in megapixels (MP). All pricing here is normalized to 1MP to allow simple comparison.
| Model | Unit | Price | Output per $1 | |
|---|---|---|---|---|
| FLUX.1 [dev] | megapixel | $0.025 | 40 megapixels | |
| FLUX.1 [schnell] | megapixel | $0.003 | 333 megapixels | |
| FLUX.1 [pro] | megapixel | $0.05 | 20 megapixels | |
| Stable Diffusion 3 - Medium | image | $0.035 | 29 images |
Output is based on 1MP images. Higher resolutions will be priced proportionally. All models listed here follow output-based pricing. Some other models may use GPU-based pricing depending on architecture.