GPU Cloud Price & Availability Tracker

Compare real-time GPU rental prices across cloud providers in one run. Get normalized, per-GPU-hour pricing for H100, H200, B200, A100, L40S, RTX 4090 and more — from Vast.ai, RunPod, and Lambda — as a clean dataset you can pipe into spreadsheets, dashboards, n8n/Make workflows, or AI agents via MCP.

What does this actor do?

Renting GPUs means checking half a dozen provider consoles, each with different units (per GPU? per node? per month?), different tiers (secure vs community, spot vs on-demand), and constantly moving prices. This actor does the comparison shopping for you:

• ✅ Queries multiple GPU clouds in parallel
• ✅ Normalizes everything to USD per GPU-hour so offers are actually comparable
• ✅ Filters by GPU model, minimum VRAM, max price, and availability
• ✅ Returns results cheapest-first, with a SUMMARY key-value record of the cheapest offer per GPU model
• ✅ Works via web UI, API, schedules, and MCP — so Claude, Cursor, and other AI agents can ask "what's the cheapest H100 right now?" and get a real answer

Who is this for?

• ML engineers & researchers deciding where to run the next training job or inference deployment
• AI startups watching burn rate — spot vs on-demand deltas across providers change weekly
• FinOps / procurement teams tracking GPU market rates over time (schedule daily runs, get a price history for free)
• AI agents doing autonomous cost optimization via MCP

Output example

{
  "provider": "Vast.ai",
  "gpuModel": "H100 SXM",
  "gpuCount": 8,
  "vramGb": 80,
  "pricePerGpuHourUsd": 1.87,
  "priceTotalHourUsd": 14.96,
  "availability": "available",
  "region": "US-East",
  "billingModel": "on-demand",
  "sourceUrl": "https://cloud.vast.ai/create/",
  "scrapedAt": "2026-07-14T15:04:05.000Z"
}

Input options

Field	Description	Default
providers	Which clouds to query	all supported
gpuModels	Filter, e.g. ["H100", "A100"]	[] (all)
minVramGb	Minimum VRAM per GPU	0
maxPricePerGpuHour	Price ceiling in USD	0 (none)
onlyAvailable	Skip sold-out capacity	true
maxResultsPerProvider	Cap per provider, cheapest first	50

Track prices over time

Schedule this actor daily (Apify → Schedules), and each run appends a timestamped snapshot. Within a month you have a GPU price history across providers — data that's genuinely hard to find anywhere else.

FAQ

How current is the data? Every run fetches live data at execution time; nothing is cached between runs.

Why is per-GPU-hour pricing the unit? It's the only unit that makes an 8×H100 node on one provider comparable to a 1×H100 instance on another.

Can I request another provider? Yes — open an issue on the actor page. Providers with public pricing APIs get added fastest.

---

Built by an AI-infrastructure engineer who got tired of tab-hopping between GPU consoles.