Pricing based on what you use

Exact pricing depends on your setup. Use the Playground to explore any configuration and instantly see the exact cost for your image or video workflow. Example costs below.

Model Size Steps Price / Video
FLUX.1 [schnell] 1024x1024 4 $0.0013
FLUX.1 [dev] 1024x1024 28 $0.0038
FLUX.1 Kontext [dev] 1024x1024 28 $0.0105
FLUX.1 Kontext [pro] 1024x1024 - $0.04
FLUX.1 Kontext [max] 1024x1024 - $0.08
HiDream-I1 Fast 1024x1024 16 $0.0032
HiDream-I1 Dev 1024x1024 28 $0.0045
HiDream-I1 Full 1024x1024 30 $0.009
SD 3 1024x1024 30 $0.0019
SDXL 1.0 1024x1024 30 $0.0019
SD 1.5 512x512 30 $0.0006
Model Resolution Duration Price / Video
MiniMax Hailuo 02 768p 10s $0.560
MiniMax Hailuo 02 1080p 6s $0.490
Seedance 1.0 Lite 720p 5s $0.143
Seedance 1.0 Lite 1080p 5s $0.151
Kling 2.1 Master 1080p 5s $0.924
MiniMax 01 Director/Live 720p 5s $0.280
Vidu Q1 I2V 1080p 5s $0.275
Kling 2.1 Pro 1080p 5s $0.323
Kling 2.1 Standard 1080p 5s $0.185
Kling 2.0 Master 1080p 5s $0.924
PixVerse v3.5 / v4 / v4.5 1080p 5s $0.299
Google Veo 3 720p 8s $4.000
Google Veo 3 + Audio 720p 8s $6.000
Kling 1.6 Pro 1080p 5s $0.323
Kling 1.6 Standard 720p 5s $0.185
Google Veo 2 720p 5s $2.500
Vidu 2.0 720p 4s $0.110
Vidu 2.0 720p 8s $0.275
Model Size Price / Video
Background Removal 1024x1024 $0.0006
Upscaling 2x 1024x1024 $0.0019
Upscaling 4x 1024x1024 $0.0019
ControlNet Preprocessing 1024x1024 $0.0006 / $0.0032
Looking for volume based discounts?

Get up to $250K transfer bonus!

Big volumes elsewhere? Runware offers the best rates in the industry, plus a monthly bonus in free platform credits, based on your current spend.

FAQ

What is Runware and what can I do with it?

Runware is a high-performance AI inference platform for generating images and videos at scale. You can use our API to create media from text, images, or video inputs using hundreds of thousands of models — instantly and affordably.

How fast is Runware compared to other platforms?

Thanks to our Sonic Inference Engine® and custom hardware, we offer industry-leading speeds: LoRA cold starts in 0.1s, checkpoint loading in 0.5s, and image/video generation in under a second. We outperform traditional cloud GPUs by 20x or more.

Is there a free trial or credit to start with?

Yes — you can generate around 1000 images on us to test the platform with no commitment.

How is pricing calculated?

Pricing depends on the model, resolution, and duration of your media. Use the Playground to test different configurations and see exact costs in real time.

Do I need to commit to a monthly plan?

No. Runware operates on a pay-as-you-go model. There are no long-term contracts or minimum commitments.

Can I get volume discounts or credits?

Yes. If you're generating media at scale, we offer up to $250K in bonus credits based on your monthly spend. Contact us to discuss volume pricing.

Why is Runware cheaper than other providers?

We’ve optimized everything — from custom hardware to green energy and software orchestration — to cut costs. Our energy-efficient infrastructure means your usage costs less than just the electricity alone on other platforms.

What kind of image generation can I do with the API?

You can generate images from text prompts, apply ControlNet guidance, use LoRAs, masks, prompt weighting, and more. Everything is configurable — width, height, steps, seed, model version, and more.

Can I use models from the Stable Diffusion ecosystem?

Yes. We support 300K+ models including custom checkpoints, LoRAs, ControlNets, VAEs, and embeddings. You can use any of them instantly via our API.

How do I know which model to use?

Our Playground lets you test any model before deploying. You can filter, preview results, and adjust parameters to find the best fit for your project.

What video models are available?

We offer top-tier models like Kling 2.1, Pixverse, Seedance, Veo 3, and more — all accessible via one unified API. You can generate high-quality short videos from text, images, or video inputs.

What are the generation modes for video?

You can generate video from text (T2V), image (I2V), or video (V2V). Advanced modes include lip-sync, multi-shot, Motion Brush, and reference-based consistency.

How long can the videos be?

It depends on the model — Kling 2.1 supports up to 5s, while others go up to 10s or more. Check each model’s capabilities in the Playground.

Is lip-sync supported in video generation?

Yes. Models like Kling 2.1 support lip-sync, motion guidance, and first-frame anchoring for better narrative control.

Can I upload my own models?

Yes. You can upload custom checkpoints, LoRAs, ControlNets, and VAEs. They’re automatically optimized for our engine and ready to use instantly.

Are my uploaded models private?

You can choose. Models can be private to your organization or made public for community access.

How do I use custom models in the API?

Each uploaded model gets a unique AIR ID. You can reference this in your API request just like any built-in model.

Can I earn money from my models?

Yes. You can license your models through our platform and earn revenue from every API call. You control pricing and terms.

Will Runware scale with my traffic?

Absolutely. Our architecture auto-scales instantly to match demand, and our parallel processing ensures sub-second generation even at peak usage.

How can I test before integrating?

Use our interactive Playground to experiment with models and configurations. It’s live, real-time, and shows exact output and cost.

How hard is it to integrate the API?

It’s designed for developers. The API is well-documented and supports straightforward JSON-based requests. No complex setup or infrastructure is required.