Salad

Salad is a very low-cost cloud option for running AI/ML production models at scale. The service taps into thousands of consumer GPUs around the world, letting customers scale up without the expense of traditional cloud computing and hyperscale services.

Among Salad's features are:

• Low cost: GPUs start at $0.02 per hour, which can cut cloud costs by as much as 90% compared with the big cloud providers.
• Scalability: Thousands of GPU instances across the globe, a good option for big AI/ML workloads.
• Fully-managed container service: No need to worry about VMs or individual machines, with a pay-by-the-minute pricing model.
• Global edge network: Low-latency nodes in many locations, good for distributed computing jobs.
• On-demand elasticity: Scale up or down quickly with Salad's toolkits or custom K8s integrations.
• GPU-accelerated processing: Can run data batch jobs, HPC workloads and rendering queues on thousands of 3D accelerated GPUs.
• Multi-cloud support: Can run alongside existing hybrid or multicloud environments.
• Security and availability: SOC2 certified, with redundant security and high availability nodes.

Salad is good for GPU-accelerated workloads, including:

• Text-to-image: Can run models like Stable Diffusion and SDXL for more images per dollar.
• Text-to-speech: Can run models like OpenVoice and Bark TTS for more inferences per dollar.
• Speech-to-text: Good for AI transcription, translation and captioning, and cheaper than traditional cloud computing.
• Computer vision: Good for running models like YOLOv8, and cheaper than traditional cloud computing.
• Language models: Good for running large language models, or LLMs, and competitive pricing.

Salad is designed to be easy to use, so customers can run models without major changes to their workflow. It works with common container registries and supports common tooling and Kubernetes workflows.

Pricing starts at $0.02 per hour for GTX 1650 GPUs, with more powerful options available. Discounts apply for large-scale deployments and monthly subscriptions. The service offers a free trial, and customers can request a personalized demo to discuss their needs and pricing.

For businesses that want to run AI/ML workloads in the cloud without breaking the bank, Salad is a good option. It offers a scalable, low-cost foundation for a variety of GPU-accelerated workloads.

Published on June 26, 2024