One GPU,
eight GPUs,
or a rack of 72.

Python

import fantasti

fx = fantasti.Client()                          # reads FANTASTI_API_KEY

gpu = fx.instances.create(
    name="ft-01",
    gpu="H100:1",                               # or "B200:8" for a whole board
    capacity=fantasti.OnDemand(),               # or Spot(...), or Reserved("rsv_…")
    volumes={"/data": "vol_datasets"},
)
cpu = fx.instances.create(name="prep-01", cpu=32)   # 32 vCPU · 128 GiB

gpu.wait("ready")
print(gpu.placement)                            # region, fabric, rate, capacity mode

region     us-east
fabric     b
rate       catalog rate · per GPU-hour
capacity   on-demand
id         ins_8f2a…

CLI

$ fantasti instance create ft-01 --gpu H100:1 --on-demand
$ fantasti instance create prep-01 --cpu 32
$ fantasti instance list

NAME      SHAPE      CAPACITY    PLACED                STATE
ft-01     H100:1     on-demand   us-east · fabric b    running
prep-01   32 vCPU    on-demand   us-east               running

REST

curl -X POST https://api.fantasti.ai/v1/instances \
  -H "Authorization: Bearer $FANTASTI_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{ "name": "ft-01", "gpu": "H100:1", "capacity": { "mode": "on_demand" } }'
# { "id": "ins_8f2a…", "status": "placing" }