Runware Compute
Reserved GPU capacity on Runware's Sonic Pods as bare-metal nodes or full clusters, or as serverless GPU compute, in the US and Europe, with starting rates by GPU model and term.
Find alternatives to Runware Compute
What it does
- GPU cloud
Sources
- Pricing
-
Product page: "Starting rates, based on term and upfront commitment · confirmed at signing" under a list of GPU hardware with per-GPU-hour starting rates.
- Description
-
Company disclosure · 16 Nov 2026
Product page text fetched 2026-11-16 (server-rendered, full text read); page quote: "Bare-metal GPU nodes and clusters, and serverless GPU compute, on Sonic Pods across the US and Europe. Half the cost of a neocloud, powered by modular infrastructure built from the ground up for AI.".
- What it does
-
Company disclosure · 16 Nov 2026
Product page text fetched 2026-11-16 (server-rendered, full text read); page quote: "Bare-metal GPU nodes and clusters, and serverless GPU compute, on Sonic Pods across the US and Europe. Half the cost of a neocloud, powered by modular infrastructure built from the ground up for AI.".
- Sold within
-
Product page read 2026-11-16: "Starting rates, based on term and upfront commitment · confirmed at signing" under per-GPU-hour reserved rates by GPU model; sold on its own rate card, not inside another offering.
- Name
-
Company disclosure · 16 Nov 2026
Product page text fetched 2026-11-16 (server-rendered, full text read); page quote: "Bare-metal GPU nodes and clusters, and serverless GPU compute, on Sonic Pods across the US and Europe. Half the cost of a neocloud, powered by modular infrastructure built from the ground up for AI.".
- URL
-
Company disclosure · 16 Nov 2026
Product page text fetched 2026-11-16 (server-rendered, full text read); page quote: "Bare-metal GPU nodes and clusters, and serverless GPU compute, on Sonic Pods across the US and Europe. Half the cost of a neocloud, powered by modular infrastructure built from the ground up for AI.".
Also from Runware
-
Sonic Inference Engine Model API
The Sonic Inference Engine serves more than 400,000 image, video, audio, text and 3D models, hosted, partner, community or uploaded, through one task-based API on Runware's own hardware.
-
Runware Serverless Infrastructure service
Serverless GPU service in which a customer brings code, a container or model weights and Runware handles GPU scheduling, workers and autoscaling from zero to production capacity, billed by the second.
-
Runware Model Router API service
Router that picks a model for each request from Runware's official catalog when the request sets the model to auto, ranked by the caller's weights for quality, speed and cost and an optional price ceiling.
-
Runware LoRA Training API service
Trains a style LoRA from a customer's uploaded images and registers it to the customer's account for generation through the same Runware API, priced per training step and with the trained weights downloadable.
-
Runware Model Serving API service
Hosts a customer's own model as a dedicated API endpoint, with Runware handling hosting, scaling and operations, billed per inference.
Something wrong here? Send a correction — quote this product id: runware-compute.