Quickstart
Five minutes: account → credits → API key → first request. The same balance also powers GPU rentals, chat and the CLI.
01Create an account
ex5.ai/login — sign up with your e-mail and a 6-digit verification code. GitHub/Google sign-in works too.
02Load credits
In the console: Billing → Add credits — with USDT or USDC (TRON, Ethereum, BNB Chain, Polygon, Base), $5 minimum. Credits are prepaid; GPUs are billed per second, models per token.
03Get an API key
Settings → API keys → + Create key. The key is shown once, at creation time only.
04First request
curl
curl https://ex5.ai/v1/chat/completions \
-H "Authorization: Bearer $EX5_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "ex5-security-27b",
"messages": [{"role": "user", "content": "Merhaba!"}]
}'DROP-IN
Already on the OpenAI SDK? One change is enough: point base_url at https://ex5.ai/v1. Streaming, tool use and JSON mode all keep working.
Public pricing feed
GET https://ex5.ai/v1/public/pricing — no authentication. For every GPU model: the lowest on-demand and interruptible price per GPU-hour in USD (the same price you are billed at launch), the number of GPUs currently available and the ISO country codes of the hosts. Refreshed every 5 minutes (Cache-Control: max-age=300). Billing is per second; a null price means no offer right now.json
{
"provider": "ex5.ai",
"currency": "USD",
"unit": "GPU-hour",
"billing": "per-second, prepaid balance, no subscription",
"updatedAt": "2026-11-07T00:23:43.041Z",
"gpus": [
{
"gpu": "RTX 5090",
"vramGb": 32,
"onDemandFromUsd": 0.49,
"interruptibleFromUsd": 0.38,
"availableGpus": 46,
"countries": ["AE", "AU", "BG", "CA", "CZ", "DE"]
}
]
}Individual offers (host, location, disk, network): GET /v1/offers?gpu=RTX%205090&type=on-demand
