Change how many GPUs this project uses, turn them on or off, or add new ones. Your app picks the table with the priority it sends.
Reserved GPUs with an SLA. Tasks wait for these GPUs only — never moved elsewhere.
| GPU | Quantity | Availability | Queue | Status | Actions |
|---|---|---|---|---|---|
NVIDIA P100 P100-16GB · 16 GB · Pascal | 1 Max 5 | Available | No backlog |
For background or non-urgent tasks, served from a shared queue.
| GPU | Quantity | Availability | Queue | Status | Actions |
|---|---|---|---|---|---|
NVIDIA T4 T4-16GB · 16 GB · Turing | 2 Max 5 | Available | 1 waiting |
Your app sends one of these with every request. Each key works for this project only. Keep keys on the server (e.g. Vercel environment variables), never in browser code.
| Name | Key | Status | Created | Last used | Actions |
|---|---|---|---|---|---|
| Vercel production | zc_live_••••c3e8 | Active | Sep 11, 2026 | 33s ago |
Three steps: add the key to your hosting, send a task, then read the result.
ZIPPA_API_URL=https://api.zippacompute.com
ZIPPA_API_KEY=zc_live_your_secret_keyconst API = process.env.ZIPPA_API_URL;
const headers = {
Authorization: `Bearer ${process.env.ZIPPA_API_KEY}`,
"Content-Type": "application/json",
};
// Your frontend calls this route; the secret key never reaches the browser.
export async function POST(req: Request) {
const input = await req.json();
// 1. Send the task to your "high" priority GPUs.
// Optional: add gpu: "T4-16GB" to pin one GPU model from that table.
const res = await fetch(`${API}/v1/jobs`, {
method: "POST",
headers,
body: JSON.stringify({ priority: "high", input }),
});
const job = await res.json(); // { id, status: "queued", queue_position: 1 }
// 2. Return the job id; the browser polls GET /api/compute?id=... for status.
return Response.json(job, { status: 202 });
}
export async function GET(req: Request) {
const id = new URL(req.url).searchParams.get("id");
const res = await fetch(`${API}/v1/jobs/${id}`, { headers });
// { status: "queued" | "running" | "completed" | "failed", queue_position, output }
return Response.json(await res.json());
}import os, time, requests
API = os.environ["ZIPPA_API_URL"]
HEADERS = {"Authorization": f"Bearer {os.environ['ZIPPA_API_KEY']}"}
job = requests.post(f"{API}/v1/jobs", headers=HEADERS,
json={"priority": "high", "input": {"prompt": "Hello"}}).json()
while job["status"] in ("queued", "running"):
print(job["status"], job.get("queue_position"))
time.sleep(1)
job = requests.get(f"{API}/v1/jobs/{job['id']}", headers=HEADERS).json()
print(job["status"], job.get("output"))API endpoint: https://api.zippacompute.com. The secret key tells us which project the task belongs to; priority picks the GPU tier (high or low). While the project is paused, requests get HTTP 423 with project_paused.
Live status your app also receives from the API.
| Task | Status | Priority | GPU · region | Submitted | Run time |
|---|---|---|---|---|---|
| job_006316 | Queued · #1 in line | Low | T4-16GB | 29s ago | — |
| job_006311 | Running | Low | T4-16GB · ap-south | 38s ago | 36s |
| job_006308 | Running | High · SLA | P100-16GB · eu-west | 43s ago | 42s |
| job_006310 | Running | Low | T4-16GB · eu-west | 59s ago | 51s |
| job_006035 | Completed | Low | T4-16GB · eu-west | 5m ago | 49s |
| job_006039 | Completed | Low | T4-16GB · eu-west | 1h ago | 53s |
| job_006031 | Completed | Low | T4-16GB · us-east | 1h ago | 47s |
| job_006048 | Completed | Low | T4-16GB · sa-east | 5h ago | 40s |
| job_006032 | Completed | High · SLA | P100-16GB · ap-southeast | 5h ago | 47s |
| job_006037 | Completed | Low | T4-16GB · eu-west | 5h ago | 32s |
| job_006033 | Completed | High · SLA | P100-16GB · us-west | 5h ago | 51s |
| job_006044 | Completed | Low | T4-16GB · eu-west | 6h ago | 45s |
| job_006028 | Completed | Low | T4-16GB · us-east | 8h ago | 1m 8s |
| job_006036 | Completed | Low | T4-16GB · ap-northeast | 8h ago | 47s |
| job_006047 | Completed | Low | T4-16GB · ap-south | 8h ago | 44s |