Provider pricing

Salad AI API pricing

Salad runs the world's largest distributed GPU network, sourcing capacity from millions of consumer PCs running RTX 3090, 4090, and 5090 cards. Pricing starts at $0.02/hour for entry GPUs and tops out around $0.30/hour for RTX 5090s — roughly 90% cheaper than hyperscalers for stateless inference and image/video generation.

Founded2019
HeadquartersSalt Lake City, UT
Tracked modelsDirectory

Best for

High-volume stateless inference workloads where cost dominates over latency SLAs.

Strengths

  • Cheapest GPU-hours on the market — RTX 4090 from $0.18/hour
  • Container-native — deploy any Docker image with autoscaling
  • Best for stateless inference: Stable Diffusion, Whisper, LLMs under 70B
  • Built-in queue, S3-compatible storage, and global low-latency routing