One platform. Multiple ways to deploy.
Salad Container Engine is the orchestration layer. Community Cloud, Salad Dedicated, and Private Cloud are the infrastructure pools it runs across. AI Gateway packages that platform behind a managed token API.
Choose the interface and compute pool that fit.
AI Gateway
An OpenAI-compatible endpoint for open-weight models. Point any SDK, agent framework, or coding assistant at ai.salad.cloud/v1 and pay per token or request a flat monthly plan. Your data is never used for training.
Salad Container Engine
The managed software layer for GPU workloads. Bring a Docker image, hardware requirements, and a replica count; SCE schedules, monitors, and reallocates across the selected infrastructure pool.
Community Cloud
The distributed pool of idle consumer GPUs that SCE deploys onto: RTX 3060 to RTX 5090, capacity across 190+ countries, and priority pricing from $0.015 per GPU-hour.
Salad Dedicated
Reserved bare metal servers and virtual machines on contract, for teams that need predictable capacity, larger VRAM, and data-center placement. RTX PRO 6000 Blackwell is the first GPU, with thousands of cards coming online.
Private Cloud
Deploy Salad's orchestration stack inside your organization and pool the idle GPUs in the workstations and servers you already own into a private, OpenAI-compatible token service. Company data never leaves company devices.
Enterprise
Deployment options, security posture, procurement, and support for organizations running Salad at scale, whether on SaladCloud, on reserved hardware, or on your own fleet.
Enterprise →Built on the stack.
Transcription API
Speech-to-text with 95.1% benchmark accuracy from $0.16 per audio hour. Translation, diarization, summaries, and captions for one price.
Learn more →Gateway Service
Dedicated residential IP tunnels through Salad nodes in nearly 200 countries, built for streaming.
Learn more →Virtual Kubelet
Schedule Kubernetes pods onto SaladCloud as container groups with no changes to your workflow.
Learn more →Recipes & models
One-click deployments of Flux, SDXL, YOLOv8, Llama, Mistral, and more, customizable to your GPU and scale.
Learn more →Not sure which deployment model fits?
Tell us about the workload. A Salad engineer will recommend the deployment model and quote it.