Runware
Pay-as-you-go API aggregating thousands of image, video, audio and LLM models with custom inference hardware for lower per-request cost.
What it does
Runware is an AI-as-a-service API that gives developers a single endpoint to access over 400,000 image, video, audio, 3D and language models. It runs on proprietary inference hardware to cut generation costs and also offers raw GPU/CPU compute billed by the second.
How to use: Integrate Runware's API into your application using Websockets, REST API, JSON, JavaScript, or Python. Access a vast library of models or bring your own, and generate images by making API requests with specified parameters like prompts, height, width, and steps.
Core features
Single API for image, video, audio, 3D and LLM models
Standardized model addressing across hosted, partner and custom uploads
Support for LoRAs, ControlNets, VAEs and embeddings on open-source models
WebSocket and REST access with async webhook delivery
Pay-per-request billing with no infrastructure to manage
Raw serverless GPU/CPU compute for custom workloads
Best for
→Adding AI image or video generation to an app without managing infra
→Batching multi-modal generation tasks in one API call
→Running custom fine-tuned models via Model Upload
→Cutting inference costs at high generation volume
Pricing
vCPU compute
$0.016/hr
RTX PRO 6000
$1.99/hr
H100
$2.76/hr
H200
$3.18/hr
B200
$4.99/hr
Toolspool rankingby monthly traffic
Tutorials
Step-by-step: exactly how to get things done with it.