Vast.ai

What's good about Vast.ai

• Significant cost savings with supply-demand pricing
• Flexible pricing with on-demand, interruptible, and reserved options
• Real-time transparent pricing with no hidden fees
• Docker ecosystem for quick software deployment
• SOC 2 Type II certified infrastructure
• Supports various AI and deep learning workloads
• CLI, SDK, and API for automated deployment
• Prebuilt templates and a model library for one-click deployment
• Per-second billing for cost optimization

Vast.ai GPU prices

71 GPU models across 189 configurations. Rank compares Vast.ai's cheapest on-demand listing for each GPU with every provider's cheapest.

What a run costs on Vast.ai

24 hours at today's cheapest listing for each GPU, at the size it is listed.

• 1× H100 SXM for 24 hours

  At $2.28/GPU-hr

  $54.67

  $22.13 under the market median rate

• 1× A100 SXM for 24 hours

  At $0.551/GPU-hr

  $13.22

  $24.94 under the market median rate

• 2× H200 for 24 hours

  At $3.94/GPU-hr

  $189.22

  $28.68 under the market median rate

The market median is every provider's cheapest on-demand price for that GPU.

Billing & terms

On-Demand

Guaranteed uptime with per-second billing. Best for production workloads.

Interruptible

50%+ cheaper preemptible instances. Best for fault-tolerant batch training.

Reserved

Up to 50% off with 1, 3, or 6 month commitments. Guaranteed capacity with volume discounts.

Serverless

Endpoint rates billed only for compute time, with no charge while workers are scaled to zero.

Regions

40+ data centers with global coverage including community and enterprise providers

Support

24/7 support chat in the console, email support at contact@vast.ai, Discord community, and documentation covering the CLI, SDK, and REST API

Watch out for

• Primarily focused on Linux-based Docker instances
• Performance may vary across different community providers
• Learning curve for users unfamiliar with marketplace-based pricing
• Limited GUI options compared to traditional cloud providers

Features

Real-Time GPU Pricing

Prices set by supply and demand across the platform with no list prices or hidden fees

Three Deployment Options

GPU Cloud for full control, Serverless for zero-ops inference, Clusters for large-scale training

Developer-Focused Tools

CLI, Python SDK, and REST API for programmatic GPU provisioning, installed together via a single package

Agent-Ready Provisioning

API-native workflow intended for AI agents to search, procure, and release compute autonomously

Model Library and Templates

Prebuilt templates for popular open-source models, plus custom templates any GPU can run with one click

Flexible Infrastructure

Scale from a small starting balance to 20,000 GPUs across 40+ data centers without contracts or minimums

Team Accounts

Shared team workspaces with role-based access for managing members and instances

Services

• GPU Cloud

  On-demand instances across 40+ data centers and 20,000+ GPUs

• Serverless

  Deploy models as endpoints with autoscaling to zero

• Clusters

  Dedicated multi-node GPU clusters with InfiniBand networking

Getting started

1. Step 1

   Add credit and get your API key

   Add credit and grab an API key from the console. No contracts, no sales calls.

2. Step 2

   Search GPUs

   Filter by model, VRAM, price, and availability via the console or API

3. Step 3

   Deploy

   Launch instances in seconds. Scale up or down programmatically.